- Email: [email protected]
Exploratory making: Shape, structure and motion
Exploratory making: Shape, structure and motion Laura Harrison, Chris Earl and Claudia Eckert, Faculty of Mathematics, Computing and Technology, The Open University, Walton Hall, Milton Keynes, MK7 6AA, UK Exploratory making activities can support the reasoning processes through which new designs are developed. With a focus on physical model-making of kinematic designs, this paper considers how these activities and processes can be articulated using formal generative rules. For kinematic designs where connections between parts afford relative motions, rule-based descriptions defining variable spatial relationships can both construct and transform models. Through modifying both shape and structure, spatial relations essential for achieving the motion characteristics of a kinematic design are identified. Unsuccessful modifications support discovery of limits to design changes, illuminate how designs work, and inform the generation of new design variations. The need for active manipulation of physical models, both to examine motions and to implement design changes, is highlighted. Ó 2015 Published by Elsevier Ltd.
Keywords: kinematic designs, design method, creative design, reasoning, reflective practice
I
n design, making appears to support the generation of new knowledge about both the use of tools and materials, and designs themselves (Ingold, 2009, 2013). Exploratory making can play a role in both analysis and synthesis, supporting the process of generating new designs, whilst also affording reflection and reasoning about their properties.
Corresponding author: Laura Harrison Laura.Harrison@ open.ac.uk
In kinematic design, physical models can support a direct appreciation of motions. In design development, modifications can be made directly to both the shape (of parts), and the structure (of connections between parts). Motion, as the primary property of interest, is a function of relationships between the shapes of parts and their connections in a design. These relationships can be complex, and visual examination of static descriptions, or even graphical simulations, may not adequately support a designer in understanding how parts move. For particular classes of kinematic designs, mathematical techniques can be used to predict motions (Hunt, 1978; O’Rourke, 2011; Phillips, 2007). For complex designs with many parts, these techniques may not reveal easily the effects of changes to shape and structure, and therefore it can be difficult to anticipate how design modifications may affect motions. Physical www.elsevier.com/locate/destud 0142-694X Design Studies 41 (2015) 51e78 http://dx.doi.org/10.1016/j.destud.2015.08.003 Ó 2015 Published by Elsevier Ltd.
51
model-making supports direct experience of motions through active manipulation of the model. For instance, Harrison, Earl, and Eckert (2011) describes how 3D printed physical models can afford an appreciation of the motions of a class of mechanisms (Figure 1). Exploratory making can also permit design modifications to be directly applied and tested. This paper examines exploratory making through the lens of creative computational search. Computational exploration can make it feasible to systematically consider possible design variations, and can provide insights into relationships between design properties. For designs with moving parts, those relationships of interest are between properties of shape, structure and motion. For example, the ‘twisted’ shapes of the connected elements of the kinematic designs in Figure 1 avoid collisions between the elements and afford continuous full-cycle motion. Creative exploration has been likened to a process of searching within an abstract space of possibilities. This analogy is attractive as it lends itself to the use of computation (Boden, 2004; Wiggins, 2001). In this view of creative discovery, theoretical spaces, containing both complete and partial solutions, can be explored to find new design possibilities. Boden suggests that individual creative processes enable the navigation of these spaces in distinct ways, with one method able to reach a solution that is inaccessible to another. Wiggins has further formalised this perspective, suggesting that these processes might be described using sets of rules. Yet how these theoretical exploration rules, and the abstract spaces they traverse, might meaningfully relate to the tangible world in which practical design exploration activities take place is unclear. Here we examine how a rule-based approach to creative search can support a more systematic approach to practical design exploration, and consider how the actions and reflections that support design reasoning through exploratory making might be described in broadly computational terms. In complex, situated practices, several types of activity may occur simultaneously. Sch€ on (1992) describes design as a process of action and reflection, which responds to the materials of the design situation. The nature of situated practices in other fields is also discussed by Suchman (1987) and de Certeau (1988). Their concepts of situated actions, and tactics, respectively, may be applicable when studying design exploration activities. As the processes that constitute making are inherently generative, we discuss how the activities and steps within an exploratory making process for designs with moving parts can be described using generative rules. We examine how these ‘making rules’ can help to articulate design reasoning processes, and support a more systematic exploration of a given design space, revealing underlying relationships between shape, structure and motion for a particular set of designs. Our method uses the design of a single existing object as a starting
52
Design Studies Vol 41 Part A November 2015
Figure 1 Building 3D-printed models to understand motions in overconstrained linkages (Harrison et al., 2011)
point, and we examine how generative making rules for both constructing and modifying physical models can be prescribed. These rules are then systematically varied, to explore a wider space containing design variations with similar motions to the original design. Here we report on an episode of exploratory making where we implement this method, selecting a mechanical toy that exhibits an unusual type of motion, as a starting point for exploration of related kinematic designs. We consider how this artefact’s shape, structure and motion can be described, and also how making rules, which describe actions for combining and transforming materials to recreate these properties, can construct physical copies. We then vary these making rules to create new designs, and examine their motions through manipulating physical models. We also examine how further alterations prompted by the making and exploration process can be applied directly to the models. We find that manipulation of physical models not only increases an appreciation of motions, but is also an integral part of the making process itself. For the case considered here, a process of systematic experimentation using physical models yields a set of making rules, which in addition to describing actions to transform and join materials, also describe the manipulations required during making. A making rule can then be generalised into a schema, defining a wider set of rules of a particular type. A specific subset of the set of making schemas generates a set of designs that each exhibit similar motions to the original object. Once identified through making and testing physical models, this subset can also help to explain which relationships between shape and structure are essential for preserving motions for this class of designs.
Exploratory making: Shape, structure and motion
53
1
Actions and reflections, plans and tactics
Different combinations of tools and materials afford opportunities in design for specific types of things to be made and distinct types of properties to be examined. Making supports the development of ideas through constructing and modifying tangible representations. It involves reasoning about relationships and connections between materials and shapes, whilst considering the effects of design changes on properties of interest. Experienced designers may rely on tacit knowledge and established techniques to help recognize opportunities for design changes and anticipate their effects, in order to select useful or meaningful actions (Lawson, 2004). However, in a new situation where previous experience is lacking, making and testing possible changes to designs quickly helps to build knowledge about the effects of these changes on various properties. Tool and material combinations may also be altered and modified as exploration progresses, to make new things visible. For three-dimensional designs, spatial interaction is necessary to examine physical objects in the round, and tactile examination may also be important. In contrast to automated processes such as digital fabrication, exploratory making affords physical action, sensory observation, and feedback. It presents opportunities to notice and pursue new ideas and directions not conceived at the outset. These characteristics are not unique to three-dimensional physical making and share features with sketching. Sch€ on (1992) studies how 2D sketching activities inform design reasoning, and describes design as a process of action and reflection, or a ‘conversation with the materials of the design situation’ (1992, p. 3). Actions make changes to tangible descriptions of designs. Reflection involves visual examination to recognize shapes and consider their formal and semantic properties. Reflecting on the desirability of properties as they emerge both identifies and motivates opportunities for further actions or changes (Sch€ on, 1992; Sch€ on & Wiggins, 1992). In other modes of making, observations to support reflection may require more that just visual examination. Knight and Stiny propose that making activities and associated reflection potentially involve a full spectrum of senses (Knight & Stiny, 2015). Actions and reflections may occur simultaneously, making them difficult to identify separately when observing designers at work. Studies of situated activities in other domains may also be relevant when considering design exploration. Suchman (1987) explains how plans (abstract, verbal descriptions of activities) may be used to predict and, to some extent, structure interactions in real-world situations before they occur, or as a means to report on them after they have occurred. De Certeau (1988), in describing activities and interactions occurring in everyday life, distinguishes between strategies, which are implemented in a top-down manner by those who have ownership of a given space, and tactics, which provide a contingent, time-
54
Design Studies Vol 41 Part A November 2015
based approach to operating in a space that is not owned, understood or predictable. Here we consider how these concepts can be applied to the spaces of material interaction where making and design reasoning occur. In this case, we interpret ownership of a design space to imply a familiar design situation, where the effects of actions on properties of interest can be readily anticipated. In such a situation, Suchman’s plans and de Certeau’s strategies can be aligned, and generative schemas can be deployed in an unsupervised manner to create new design variations. De Certeau’s tactics are employed when continually responding and reacting to the materials of the situation, to notice opportunities for action, exploit them, and observe the consequences. In this sense, they appear similar to Sch€ on’s actions and reflections, with one key distinction: although both strategies and tactics may operate simultaneously in a given physical space, they afford a binary distinction. For de Certeau, these are the separate modes of operation of distinct groups from different political spheres, which have little scope for translation between them. In contrast, in design exploration there exists a continuous spectrum between the new and unpredictable and the familiar and understood, where the continued application of tactics may eventually afford the development of strategies. Repeated actions and reflections support situated learning about the nature of a certain situation, developing a growing appreciation of inherent relationships between properties, and improving confidence in the reliability of a particular approach. But even when reliable sequences of actions become practised and polished by the rhythms of repeated application, how can they be described, documented, or discussed? Communicating verbally the detailed nuances of material making, even when the maker has developed a practised intimacy with the activity, and all its opportunities for error, is often not straightforward. Making seems best demonstrated through practice, in a language of tools and materials, which together afford precise motions, operations, and sensations. When described out of context, all complex situated activities must become to some extent simplified into abstract schemes, lacking the essentially unbounded contextual information that defines how activities unfold in practice. In Suchman’s outlook, plans describe the steps or stages of an activity to the extent that they can be meaningfully articulated, either in advance or as a means of reporting after the fact. Situated actions then develop a finer resolution as appropriate, in direct response to a situation. For human reasoning this requirement for finer resolution in description is no obstacle to the fluent conception and implementation of appropriate responses as circumstances present themselves. This presents a key conceptual distinction between situated actions and exploratory tactics: practised situated actions, although
Exploratory making: Shape, structure and motion
55
they cannot be articulated easily in full detail, are routinely relied upon to deliver successful results, whereas with exploratory tactics, outcomes are not necessarily predictable. This study focuses on describing and documenting moves and actions by which designers explore and develop understanding in unfamiliar territories of making. We use the example of kinematic designs, where technical and conceptual difficulties in predicting outcomes call for a more effective, exploratory making approach.
2
Generative design exploration
Generative design descriptions use rules to define sets of transformations. When applied to tangible media or materials, generative rules define ‘instruction sets’ for processes to synthesise tangible things. In exploratory making, synthesising tangible designs, that is, physical artefacts, and then observing and evaluating their properties directly, provides a measure of separation between action and reflection, whilst retaining continuity in the process of design development. Generative descriptions alone may not afford an appreciation of the properties of the designs they describe, but they can make it easier to make observations about the transformation processes that construct them, facilitating discussions about design processes.
2.1
Shape rules and schemas
As shape descriptions of designs can, by definition (Stiny, 1981), allow division into parts in infinitely many ways, activities such as sketching that support the development of unstructured shape descriptions allow designs to be continually reinterpreted. As shapes are added or selected and transformed, they may interact or merge with other shapes, in ways not predictable in advance. Shape rules can be used to describe and document the distinct moves by which shapes within designs are recognised and transformed: rules identify spatial relationships between shapes and apply transformations to shapes and subshapes within designs (Stiny, 2006). Shape grammars employ sets of shape rules to construct complete designs, and can describe sets of designs that exhibit distinct visual styles or properties (Flemming, 1987; Koning & Eizenberg, 1981; Prats, Earl, Garner, & Jowers, 2006; Stiny & Mitchell, 1978). Shape schemas provide general descriptions of classes of rules, which can be instantiated in many ways, to create specific rules for particular applications. Stiny (2011) identifies several levels of general schemas, classifying types of shape rules, including for example parametric variations, and geometric transformations such as rotations, reflections, and translations. Schemas can also be specified for joining and dividing parts, selecting shape boundaries, or changing dimensions of representation by creating shapes from their boundaries and vice versa. The identity schema provides a special case, because it
56
Design Studies Vol 41 Part A November 2015
recognises shapes but does not apply further transformations (Stiny, 2011, 2006). These ‘useless’ rules at first glance do not appear to do anything, because they do not visibly transform designs. However, for structuring shape descriptions, these rules permit new shapes to be noticed or identified (Stiny, 1996). Shape rules have been used widely for the analysis of existing designs (e.g. Flemming, 1987; Koning & Eizenberg, 1981; Stiny & Mitchell, 1978). By considering the spatial relationships encountered within these designs, underlying rules can be inferred. Sets of such rules define grammars that can be used both to reconstruct existing designs and to discover new variants. Knight and Stiny (2001) also advocate an alternative, more open-ended use. Shape rules support systematic yet creative exploration and can be employed in an open-ended way (Knight, 1992, 1995, 2005; Stiny, 1980).
2.2
A method for exploratory making
We outline a method for using generative rules in a design exploration process, for the example of kinematic designs (that is, shapes with moving parts). In contrast to employing sequential actions and reflections to develop a design along a single trajectory, we consider how systematic exploration, from an existing design as a starting point, can uncover other related designs that exhibit similar characteristics e in our example, similar motions. Through experimentally varying shape and structure in physical models, our aim is to discover which relationships in an original design are essential to preserving motions, and conversely which design variables might be freely varied as required by a functional application. First, we define a set of making rules, which describe the actions and manipulations required to fabricate a working copy of an existing mechanical artefact. Systematically varying these making rules, through abstracting them into a more general set of making schemas, yields a set of physical models that embody a wide range of variations upon the original design. Evaluating the properties of each, through manipulating and observing motions, develops a more precise understanding of the scope for transformations within design constraints on kinematic behaviour. Identifying particular subsets of making schemas that reliably reproduce the motions of the original artefact helps to explain how a particular design works.
2.2.1
Selecting a starting point
Our approach employs an existing kinematic design as a starting point. Designs appropriate for the method are those that are not already well-known and well-explored by conventional analysis techniques. Kinematic designs with distinctive, unusual motions are also advantageous because the presence or absence of distinctive motion characteristics will be apparent immediately. To avoid explorations becoming limited artificially by expectations about an
Exploratory making: Shape, structure and motion
57
object’s function, designs that have no familiar functional applications are preferred.
2.2.2
Manipulating physical models
Physical models of designs with moving parts afford an interrogation of possible motions through manipulation. Certain variable spatial relationships are inherently afforded by particular designs, but these must be elicited and observed through direct interaction, so that the relative success of a modified design in reproducing these motions can be evaluated. In some cases modified kinematic designs exhibit restricted motions (perhaps through transformations of shape causing collisions, or transformations of connections causing reduced mobility). Descriptions of the configuration states into which the physical model of the kinematic design can be manipulated provide a means to describe motion characteristics.
2.2.3
Describing a design in terms of shape and structure
The shapes of parts, and the types of the connections between them in a kinematic design, describe the shape and structure of the design. Shape rules afford an immediate route to describing shapes and connections (Stiny, 1981).
2.2.4
Making rules for motions
Certain modes of material combinations allow for the fabrication of specific connection types more readily than others. For hinged designs, for example, simple materials readily afford the construction of physical models that exhibit appropriate variable spatial relationships between parts. More sophisticated connection types may require more advanced tools, materials, and fabrication processes. Here, making rules are used to describe the actions and manipulations employed for shaping and connecting materials whilst constructing physical models.
2.2.5
Abstracting making rules into tactical schema
Removing detail from making rules for shape and structure can define making schemas, describing sets of rule variations. Particular combinations of these new rule variations can then be employed to construct physical models of new designs that are related to the original object.
2.2.6
Evaluating motions of new designs
New models can then be manipulated using the actions discovered when interacting with the original artefact. This enables evaluation of the extent to which motions are preserved in the new design.
2.2.7
Situated tactics
Whilst the original making rules provide a starting point from which to derive new rule variations and construct new designs, new physical models
58
Design Studies Vol 41 Part A November 2015
themselves may prompt additional ideas for alterations that can be applied directly to the objects.
2.2.8
Discovering strategic schemas
Through systematically constructing and considering model variations, subsets of the tactical making schema (see Section 2.2.5) that create new objects which reliably reproduce the motions of the original design are identified.
3
Rules for making
Whilst shape rules are useful for describing aspects of designs with moving parts, they are not sufficient to fully encode the types of action and reflection needed to construct and evaluate them. Knight and Stiny (2015) suggest that the concept of identity rules for recognizing shapes could be expanded to create more general ‘sensing’ rules to describe other forms of observation and reflection, and also that shape rules can provide a basis for defining different types of ‘doing’ rules for transforming materials. Sets of doing and sensing rules can be employed within making grammars, to describe making activities of particular types. Here, we consider how shape rules can be adapted to define making rules for kinematic designs. As doing and sensing can occur simultaneously, making rules must support both action and reflection. We identify two specific types of rule, which are essential for considering kinematic designs. One of these defines the creation of various types of connections between parts, which possess various classes of variable spatial relationships. The other type describes manipulative actions and interactions on models. These manipulation rules determine the overall motions afforded by the combined effects of the variable spatial relationships defined by each connection.
3.1
Describing structure and variable spatial relationships
For three-dimensional models composed of only rigid materials, joints or connections between pairs of rigid parts impose constraints on their relative motions. These connections are termed kinematic pairs. In lower kinematic pairs (Figure 2), connected parts share surfaces of contact, whereas higher pairs exhibit point or line contact (Reuleaux, 1876). Shape rules might be extended to describe subshape relations that afford variable spatial relationships between parts in kinematic pairs (Figure 4). For lower kinematic pairs, subshape relations can use surface embedding to define parts of surfaces as shared sub-shapes within both parts. Motions observed in lower kinematic pairs due to these shared surfaces can also be modelled by shape representations using sub-shapes of lower dimensions, such as shared lines and points (see Figure 3).
Exploratory making: Shape, structure and motion
59
Figure 2 Lower kinematic pairs
Figure 3 Connections that create variable spatial relationships described by shared sub-shapes
60
Design Studies Vol 41 Part A November 2015
Figure 4 A shape rule describing a variable spatial relationship
3.2
Manipulation
The potential for parts to move within a design is directly related to the composition of its connections and the shapes of its parts. Descriptions of shape and structure therefore inherently contain information about a design’s motions. However, when many variable spatial relations interact, the overall motions that result may not be straightforward to determine. Physical models, embodying three-dimensional descriptions of both shape and structure, provide an expedient way to both examine and describe tangibly the motions occurring in these designs (Video S1a and b). Interaction with unfamiliar physical objects is both tactical and situated, as manipulation itself is exploratory. However, through a process of experimentally applying forces, particular sequences of actions that successfully
Video S1a.
Exploratory making: Shape, structure and motion
61
Video S1b.
articulate a design’s full range of motions may come to be understood, and these can then be systematically applied. In further design exploration, practised manipulation sequences may be experimentally applied to new designs, to see if motion sequences encountered previously are preserved. Because manipulation rules are situated and act directly upon models, it is difficult to define them out of context. Further, as the motions they elicit from models may be complex, it may not always be possible to succinctly describe manipulations in terms of the variable spatial relations they enact. Here, rather than attempting to describe the manipulation rules required for design exploration in full detail, we consider that, from a set of all possible manipulations, a particular sub-set is discovered to successfully elicit the motions of a given artefact.
4
An example of exploratory making
We report on an episode of systematic exploration through making, implementing the method outlined in Section 2.2. This practical work was undertaken by the first author. The actions, interactions and reflections that occurred were observed and documented at every stage. In addition, as care was taken to construct a new model each time a design change or alteration was applied, the resulting collection of models provides a lasting physical record of the design activity that was undertaken.
4.1
Selecting an existing object as a starting point for design exploration
The chosen starting point was a mechanical toy that exhibits an unusual sequence of motions, which fulfil no immediate function beyond their ability
62
Design Studies Vol 41 Part A November 2015
Video S2.
to entertain. The toy is composed of two connected layers of flat, hinged panels that form a rectangular sheet (see Video S2). We encountered this object as a sample model from a Selective Laser Sintering equipment manufacturer. A 3D printable copy is available (Thingiverse, 2013). We have not yet discovered the designer of the original. However, we discovered a patent for a similar design that exhibits an identical sequence of motions, but is composed of only one layer, and intended to be folded from a single sheet of card (Byrnes, 2004). We expect that the two-layer version may be an intentional variation, in order to achieve the same motion sequence as the protected design without infringing on the shape and structural descriptions specified in the patent.
4.2
Discovering and describing motions through manipulation
The toy’s hinged panels can be manipulated into four distinct states. When manipulated continuously, the toy transforms between these states consecutively in a continuous cycle. Figure 5 illustrates the sequence of unfolding actions through which the toy eventually returns to its initial state. This type of motion sequence is referred to as full-cycle mobility (Byrnes, 2008). This toy design exhibits non-continuous motions, as symmetric pairs of hinged panels can fold through a maximum of 180 before shape interference limits motion. (In contrast, the designs in Figure 1 and Video S1a and S1b exhibit continuous full-cycle mobility.) During each unfolding action, two symmetric hinged segments are rotated until a limit to motion is reached. From this new state, new sets of hinges become aligned, and a new phase of motion becomes possible.
Exploratory making: Shape, structure and motion
63
Figure 5 Motion sequence between the four states (note that sequence may also occur in reverse)
The effects of the manipulation sequence upon shape and structure can be described by examining the toy in its four possible states from different perspectives, and then creating drawings as shown in Figure 6. This non-continuous behaviour is relatively uncommon. The nature of these motions was unanticipated upon first encounter with the object, but became apparent to us, and to several others informally presented with the artefact, after some initial exploratory manipulation. Practice then rendered familiar the sequence of actions required to cycle through states continuously. Within the following exploration exercise, this sequence of actions was used to test whether the new physical models created also exhibit similar motions.
4.3
Describing spatial relationships between shape and structure
Whilst considering all reachable states or configurations helped to describe and examine motion, the toy’s shape and structure (that is, its parts and connections) alone could be represented just by the appearance of the toy in State 1. In this state, the toy resembles a flat sheet (Figure 7). Drawings of each face in this configuration were relatively simple to produce, and embody a large number of particular spatial relationships (see Table 1, column A).
64
Design Studies Vol 41 Part A November 2015
Figure 6 Drawings of the toy design in its four distinct states
4.4
Rules for making
From this description of shape and structure, we developed making rules for each layer. To construct a working copy of the design, common modelmaking materials that permitted rapid construction and modification of new models were employed. Because the design is composed of flat, hinged panels,
Figure 7 Spatial relationships in the original design
Exploratory making: Shape, structure and motion
65
Table 1 Spatial relationships in (A) the original design, (B) as simplified after testing the effects of rule variations on motions, and (C) the essential relationships discovered through exploration (in bold) with other related descriptions that support systematic making (in parentheses)
A. Initial spatial relationships
B. Related lesser statements
C. Essential relationships (and related lesser descriptions to support making)
The overall shape of the design is a The overall shape is a rectangle. rectangular sheet, twice as long as wide. The edges of the overall shape are straight.
(The overall shape may be any closed curve that contains the hinge lines and their intersections.)
The hinge-lines on a given face are parallel to the edges of the sheet, and therefore also to each other.
Hinge-lines within pairs on each face must be parallel to each other.
Hinge-lines within pairs on each face must be parallel to each other.
The hinge-lines are positioned 1/4 and Pairs of hinge-lines are positioned (Interfering material may need to 3/4 along the face. symmetrically about a centre-line. be tactically removed.) The hinge-lines are spaced apart by half the panel’s full width/length. Hinge pair orientation and spacing is identical but perpendicular on each face.
Pairs of hinge-lines must be Pairs of hinge-lines must be perpendicular to those on the opposing perpendicular to those on the face. opposing face.
The split-line must be perpendicular to The split-line is a straight line. the hinge-lines. The split line or curve on one face is perpendicular to the split line on the opposing face. The design is composed of 2 full sheets of material.
(The split-line or curve on a given face must bisect both hinge lines on that face.) (Enough material must be present to maintain the object’s integrity and support attaching the two layers together.)
we used foam-board to cut 2D shapes, and tape to selectively re-join panels along cut-lines to instantiate hinges. Shape rules, with conventions for describing hinge-lines, could then describe how to shape and join materials (Figure 8). Whilst these do not describe the full details of the complex practical activities of cutting and joining, this contextual knowledge could be accessed readily in the situated application of schematic rules for making. External examination alone, however, did not identify how the two panel layers, assembled using Rules 1, 2 and 3, should be joined together. Situated experimentation yielded this information. Manipulating the two assembled sections established which parts must move and therefore should not be attached. This manipulation was conducted in parallel with the tentative application of adhesive at various points to test whether the model would still transform as expected. Figure 9 demonstrates schematically a reasoning process that determines a rule for where adhesive can be applied. We observed that tactics of trying and testing alone discovered the same result. Figure 10 describes this additional making rule, required for the complete assembly of a working
66
Design Studies Vol 41 Part A November 2015
Figure 8 Rules for making
Figure 9 A reasoning schematic to derive Rule 4 (for joining the two panel layers)
model. Manipulation through the four states confirmed that this model reproduces the motions of the original toy.
4.5
Varying making rules to give tactical schema for design exploration
More models were constructed by varying the original rules to explore and test which aspects of the original design are important for preserving motions. Note that at the outset we had no intuition as to which variations would be successful, but we attempted to explore a wide range of possible changes systematically. We considered how rules might be simplified and abstracted into more general schemas, from which new variants on the original making rules could be defined. For each variation, we constructed a new physical model. Figure 13 shows the full set of models constructed (Models 1e14). The first variation (Model 1) generalized the proportions for the rectangle in Rule 1 to allow parametric variations. When the overall shape is set to a square,
Exploratory making: Shape, structure and motion
67
Figure 10 A rule 4 for making: gluing layers
both faces become identical, and the schematics for Rules 2 and 3 become simplified (Figure 11). In this case, for assembly, one layer must be rotated through 90 relative to the other. Through allowing parametric variation, Rule 10 evolves into Schema 10 (Figure 12), which constructs both square and rectangular panels of all proportions (for example Model 14). Schema 100 is further abstracted, to allow for non-rectangular shapes (Figure 12; Figure 19). A second variation tested whether the pairs of hinge-lines created by Rule 2 must be located as precisely as originally specified, or whether a more general schema (Schema 2, Figure 12) would be acceptable. In the original design, outer edges meet precisely along a centre-line upon folding inwards. Avoiding
Figure 11 From Rule 1 to Schema 1 and simplified rules 2 and 3
68
Design Studies Vol 41 Part A November 2015
Figure 12 From making rules to making schema
overlapping and associated interference requires that the spacing between the hinge lines must be at least half the width of the face, but a variation that relaxed this rule so that hinges might also be further apart, was also tested (Models 7 & 8). Further variations tested whether hinge-lines must be symmetrically placed, deriving Schema 20 (Figure 12; Models 13 and 14), whether hinge lines must lie parallel to the panel edges (Schema 200 , Figure 14; Model 4), or parallel to each other (Model 12). When varying hinge alignment, external panel edges can meet at an altered centre-line upon folding, and the greatest possible deviation would orient this centre-line along the diagonal of 45 . To achieve this, hinge lines were rotated by half this angle (Models 4, 5, 10, 11). To preserve the perpendicularity of hinge alignment between the faces during this variation, the layout of one face must be a mirror image of the other, rotated by 90 , rather than an absolute copy (Model 9 tested whether orientation between the 2 layers could also be varied). Further variations (Figure 12) tested how Rule 30 might be simplified: by removing the requirement that the split-line be perpendicular to the hingelines on that face (Models 4 and onwards; Schema 3); removing the requirement that it must be a straight line (Schema 30 ; Models 13 and 14); and also testing whether split-lines must be placed similarly on both faces (Models 13 and 14).
4.6
Evaluating motions of new models through manipulation
After fabrication, each new model was manipulated using the same sequence of actions as was discovered to elicit the motions of the original toy (see Section 4.2). When a new model could be manipulated successfully through all four states sequentially, the rule variations that constructed it were deemed to be allowable for preserving those spatial relationships essential for achieving the motions exhibited by the original design. Table 1 outlines how
Exploratory making: Shape, structure and motion
69
Figure 13 Experimental models to test design variations
70
Design Studies Vol 41 Part A November 2015
Figure 14 Schema 5 illustrates a tactic (consequent on Rule 200 ) to remove and reattach interfering material. The notation M denotes a set of manipulations
this new knowledge of successful variations allows the descriptions of spatial relationships observed in the original design to be relaxed, creating less precise descriptions that define wider categories of objects, and ultimately identifying essential relationships for achieving motions.
4.7
Applying situated tactics to directly modify objects
In Section 4.4, we noted the need for situated, tactical experimentation in order to derive Rule 4. This need for a situated reaction to models, to conceive of new making rules, continued to occur throughout our process of exploration. This was particularly important when rule variations initially appeared to create unsuccessful models, which could not be manipulated through the full four-state cycle. When Schema 200 was applied (see Figure 14) the resulting model (Model 4a) could be manipulated into states 2 and 4, but material interference prevented state 3 being reached. However, here a new idea for a situated rule was identified: when examining the design in states 2 and 4, its footprint was noticed to be larger than its square shape boundary in State 1. A tactic of trimming sections that protruded beyond the original square footprint was tried. This resulted in a model that could then reach all four states successfully (Model 4b).
Exploratory making: Shape, structure and motion
71
It was also noticed later that this procedure could be further refined, and slightly less material could be removed, while still yielding a working design (Model 10). Interfering material could be marked precisely, directly on the model, in states 2 and 4 respectively, and these marked sections could then be removed. It was found latterly that this trimming approach could be applied successfully to complete any design variation where material interference was initially encountered. Because this occurred for a significant subset of the designs produced here, this new schema, Schema 5 (Figure 14), became an important addition to the making sequence. Although conceived through situated experimentation, once practised and found to be a reliable method, this schema could be defined precisely and applied systematically. Note that the original making rules assemble the design exclusively in State 1. However, to discover and remove interfering material, Schema 5 relies on subsequent manipulations, using the set of manipulations M to transform the model into and between states 2 and 4 (see Figures 5 & 6), so that marking and trimming operations can be applied. Therefore actions for manipulating designs between states (see Section 4.2) are required not only for testing the success of designs, but also as part of the making process itself. Further situated tactics were uncovered by considering how successful models might be altered further, by selectively removing material. Rule 4 (Figure 10) indicated that material that lies outside of the hinge-lines plays a role in connectivity between the two layers, but material lying inside the layers might be less essential. Through experimentally removing this material, we ascertained that this was indeed the case (Model 2) and defined a corresponding Schema 6 (Figure 15). Further, considering the resulting model in its intermediate states indicated that further portions of material also played no role in essential connectivity. Also removing these parts led to a pared-down model (Model 3), which changes its footprint dramatically upon transforming between states (Figure 16). These reducing rules were found to also be applicable to other model variants, producing a subset of designs that possess the additional property of expanding and collapsing between states (Models 5, 7, 8). These pareddown designs could also be constructed from a single sheet of thin material. They therefore appear to fall within the subset of designs protected by the patent. Note that the resulting designs do not require full sheets of material in their construction. Thus, they could also be constructed more efficiently from their necessary parts directly using Schema 7 (Figure 17).
4.8
Strategic schemas for making
After making a number of models, we established that many of the relationships inherent in the original design were inessential for preserving its motions.
72
Design Studies Vol 41 Part A November 2015
Figure 15 Schema 6 e removing material
Figure 16 The two pared-down models (3 and 5), in states 1e4 with expanding and collapsing footprint
Figure 17 Schema 7 to assemble wholes from parts
Exploratory making: Shape, structure and motion
73
Significantly, the only relationships that proved essential were the relative orientations of the hinge-lines, which define a particular set of variable spatial relationships between the design’s parts (see Table 1, column C: bold text). Models 9 and 12 (Figure 13), in which hinge orientations were altered, were found to be unsuccessful, not due to material interference, which can be fixed using Schema 5 (Figure 14), but due to structural misalignment. This apparent importance of hinge orientation indicated that Schema 2 (Figures 12 and 14), which defines hinge geometries, is perhaps the most critical step in the making process. Because other design aspects are constrained only by their locations relative to these hinge lines, placing Schema 2 at the beginning of the making sequence defines a more strategic approach to constructing designs. An evolved Schema 2* (see Figure 18) initially defines relations between hinges for both layers. As the remaining panel edges may otherwise form any closed shape, a corresponding Schema 1* must merely place material outside the hinges and their intersections to ensure practical construction. A Schema 3* adds a split-line or curve to each face, bisecting both hinges. As before, Schema 4 is required for gluing, and Schema 5 is sometimes necessary for removing material interference. These new, strategic schemas, outlined pictorially in Figure 18, permit a more direct, strategic approach to constructing further designs. With the allowable limits to variations of shape and structure now clearly defined, creative exploration can focus on other design properties, such as colour and composition (Figure 19). Formally defining the manipulations and alterations that occur within Schema 5 can also support further creative exploration using digital drawing tools. The interplay between exploration in physical making and in other representations, including digital tools, is the subject of further work by the authors.
5
Discussion
This exercise in exploratory making goes some way towards demonstrating a possible role of a generative approach in developing understanding of relationships between shape, structure, and motion in kinematic designs. For the case of the single design considered here, practical design reasoning using exploratory model-making successfully derived sets of making schemas that constructed larger sets of designs with similar properties. Further, these schemas also possess an explanatory function, illuminating how or why the original design works. Significantly, models that embodied unsuccessful changes were often more useful for positively identifying essential relationships than successful models.
74
Design Studies Vol 41 Part A November 2015
Figure 18 Schemas for strategic making
Usually seen as abortive experiments, these unsuccessful results of design exploration played a critical role in both identifying and explaining essential relationships. In our study, alterations that affected spatial relationships between hinge-lines were discovered to exceed allowable limits for variations. This discovery assisted the positive recognition of boundaries within which motions are preserved, thereby allowing reliable generative schemas to be developed. These schemas, in turn, helped to explain which aspects of shape and structure found in the original design were important for preserving its motions in new design variations. The design we began with possessed unusual and distinctive motions, and therefore the presence or absence of these full-cycle sequences in new design models could be evaluated readily. A single example has been considered here, but further work to consider other classes of designs could establish the extent to which this approach might be applied more widely. We expect that the method is transferable most easily to types of kinematic design composed of closed linkages, where the overall motion of the design is highly constrained. In our example, unsuccessful designs were those where full-cycle motion sequences were no longer possible. The binary nature of this test makes it far easier to apply than detecting subtler changes in more complex, or lessconstrained, kinematic behaviours. Techniques and language for describing detailed motions in designs directly are not commonly available, and description in this paper focused primarily on shape and structure. We note that manipulation of physical models to
Exploratory making: Shape, structure and motion
75
Figure 19 Creative designs developed using the evolved making schemas
examine motions was at first tactical and exploratory in nature, but later a practised sequence of manipulations became familiar and routine. This example highlights the importance of physical manipulation when considering kinematic designs with moving parts. Manipulation develops an understanding of motions, tests the success of design modifications, and also plays a role within the synthetic steps of the making process. Physical manipulations and associated exploratory making are a situated set of actions and evaluations. Situated behaviours in other fields have helped to provide a framework for considering the tactical manner in which appropriate manipulation sequences are discovered through material interaction.
76
Design Studies Vol 41 Part A November 2015
Because anticipating effects of changes to shape and structure is difficult in the kinematic design we considered, the tactical nature of exploration assumes a particularly important role. The importance of tangible models, for affording opportunities and ideas for new transformations and modifications, also became apparent, as these new directions could not be readily conceived through considering abstract descriptions alone. Further, our collection of experimental physical models also seems to play a useful role in presenting this work to others. As tangible representations of motions, they provide an expedient route for disseminating the explanations about relationships between shape, structure and motion that were uncovered by our generative approach.
References Boden, M. A. (2004). The Creative Mind: Myths and Mechanisms. London: Routledge. Byrnes, R. B. (2004). Playthings. U.S. Patent No. 20,040,092,199. Washington, DC: U.S. Patent and Trademark Office. Byrnes, R. (2008). Metamorphs: Transforming Mathematical Surprises. Tarquin Publications. de Certeau, M. (1988). L’invention du quotidien. In: Arts de faire, Vol. 1. Berkeley, CA: University of California Press. Flemming, U. (1987). More than the sum of parts: the grammar of Queen Anne houses. Environment and Planning B: Planning and Design, 14(3), 323e350. Harrison, L. C. A., Earl, C. F., & Eckert, C. M. (2011). The role of rapid prototyping in the generative design of mechanisms. In 12th Conference on Rapid Design, Prototyping & Manufacturing (RDPM 2011). Hunt, K. H. (1978). Kinematic Geometry of Mechanisms. London: Oxford University Press. Ingold, T. (2009). The textility of making. Cambridge Journal of Economics, 34(1), 91e102. Ingold, T. (2013). Making: Anthropology, Archaeology, Art and Architecture. London: Routledge. Knight, T. W. (1992). Designing with grammars. CAAD Futures’ 91: Computer Aided Architectural Design Futures: Education, Research, Applications 33e48. Knight, T. W. (1995). Transformations in Design: A Formal Approach to Stylistic Change and Innovation in the Visual Arts. New York, NY: Cambridge University Press. Knight, T. (2005). Creativity rules. In J. S. Gero, & M. L. Maher (Eds.), Computational and Cognitive Models of Creative Design VI (pp. 155e174). University of Sydney: Key Centre of Design Computing and Cognition. Knight, T., & Stiny, G. (2001). Classical and non-classical computation. Arq: Architectural Research Quarterly, 5(04), 355e372. Knight, T., & Stiny, G. (2015). Making grammars: from computing with shapes to computing with things. Design Studies, 41(PA), 8e28. Koning, H., & Eizenberg, J. (1981). The language of the prairie: Frank Lloyd Wright’s prairie houses. Environment and Planning B, 8(3s), 295e323. Lawson, B. (2004). Schemata, gambits and precedent: some factors in design expertise. Design Studies, 25(5), 443e457.
Exploratory making: Shape, structure and motion
77
O’Rourke, J. (2011). How to Fold It: The Mathematics of Linkages, Origami, and Polyhedra. Cambridge University Press. Phillips, J. (2007). Freedom in Machinery. New York, NY: Cambridge University Press. Prats, M., Earl, C., Garner, S., & Jowers, I. (2006). Shape exploration of designs in a style: toward generation of product designs. AIEDAM, 20(03), 201e215. Reuleaux, F. (1876). The Kinematics of Machinery: Outlines of a Theory of Machines. London: Macmillan & Co. Sch€ on, D. A. (1992). Designing as reflective conversation with the materials of a design situation. Knowledge-Based Systems, 5(1), 3e14. Sch€ on, D. A., & Wiggins, G. (1992). Kinds of seeing and their functions in designing. Design Studies, 13(2), 135e156. Stiny, G. (1980). Kindergarten grammars: designing with Froebel’s building gifts. Environment and Planning B, 7(4), 409e462. Stiny, G. (1981). A note on the description of designs. Environment and Planning B, 8(3), 257e267. Stiny, G. (1996). Useless rules. Environment and Planning B: Planning and Design, 23(2), 235e237. Stiny, G. (2006). Shape: Talking About Seeing and Doing. Cambridge, MA: The MIT Press. Stiny, G. (2011). What rule (s) should I use? Nexus Network Journal, 13(1), 15e47. Stiny, G., & Mitchell, W. J. (1978). The Palladian grammar. Environment and Planning B, 5(1), 5e18. Suchman, L. A. (1987). Plans and Situated Actions: The Problem of HumanMachine Communication. New York, NY: Cambridge University Press. Thingiverse. (2013). Flipping Folding Fun Toy. Retrieved from. http://www.thingiverse.com/thing:53658. Wiggins, G. A. (2001). Towards a more precise characterisation of creativity in AI. Case Based Reasoning: Papers from the Workshop Programme at ICCBR, 1, 113e120.
78
Design Studies Vol 41 Part A November 2015
Keywords: kinematic designs, design method, creative design, reasoning, reflective practice
I
n design, making appears to support the generation of new knowledge about both the use of tools and materials, and designs themselves (Ingold, 2009, 2013). Exploratory making can play a role in both analysis and synthesis, supporting the process of generating new designs, whilst also affording reflection and reasoning about their properties.
Corresponding author: Laura Harrison Laura.Harrison@ open.ac.uk
In kinematic design, physical models can support a direct appreciation of motions. In design development, modifications can be made directly to both the shape (of parts), and the structure (of connections between parts). Motion, as the primary property of interest, is a function of relationships between the shapes of parts and their connections in a design. These relationships can be complex, and visual examination of static descriptions, or even graphical simulations, may not adequately support a designer in understanding how parts move. For particular classes of kinematic designs, mathematical techniques can be used to predict motions (Hunt, 1978; O’Rourke, 2011; Phillips, 2007). For complex designs with many parts, these techniques may not reveal easily the effects of changes to shape and structure, and therefore it can be difficult to anticipate how design modifications may affect motions. Physical www.elsevier.com/locate/destud 0142-694X Design Studies 41 (2015) 51e78 http://dx.doi.org/10.1016/j.destud.2015.08.003 Ó 2015 Published by Elsevier Ltd.
51
model-making supports direct experience of motions through active manipulation of the model. For instance, Harrison, Earl, and Eckert (2011) describes how 3D printed physical models can afford an appreciation of the motions of a class of mechanisms (Figure 1). Exploratory making can also permit design modifications to be directly applied and tested. This paper examines exploratory making through the lens of creative computational search. Computational exploration can make it feasible to systematically consider possible design variations, and can provide insights into relationships between design properties. For designs with moving parts, those relationships of interest are between properties of shape, structure and motion. For example, the ‘twisted’ shapes of the connected elements of the kinematic designs in Figure 1 avoid collisions between the elements and afford continuous full-cycle motion. Creative exploration has been likened to a process of searching within an abstract space of possibilities. This analogy is attractive as it lends itself to the use of computation (Boden, 2004; Wiggins, 2001). In this view of creative discovery, theoretical spaces, containing both complete and partial solutions, can be explored to find new design possibilities. Boden suggests that individual creative processes enable the navigation of these spaces in distinct ways, with one method able to reach a solution that is inaccessible to another. Wiggins has further formalised this perspective, suggesting that these processes might be described using sets of rules. Yet how these theoretical exploration rules, and the abstract spaces they traverse, might meaningfully relate to the tangible world in which practical design exploration activities take place is unclear. Here we examine how a rule-based approach to creative search can support a more systematic approach to practical design exploration, and consider how the actions and reflections that support design reasoning through exploratory making might be described in broadly computational terms. In complex, situated practices, several types of activity may occur simultaneously. Sch€ on (1992) describes design as a process of action and reflection, which responds to the materials of the design situation. The nature of situated practices in other fields is also discussed by Suchman (1987) and de Certeau (1988). Their concepts of situated actions, and tactics, respectively, may be applicable when studying design exploration activities. As the processes that constitute making are inherently generative, we discuss how the activities and steps within an exploratory making process for designs with moving parts can be described using generative rules. We examine how these ‘making rules’ can help to articulate design reasoning processes, and support a more systematic exploration of a given design space, revealing underlying relationships between shape, structure and motion for a particular set of designs. Our method uses the design of a single existing object as a starting
52
Design Studies Vol 41 Part A November 2015
Figure 1 Building 3D-printed models to understand motions in overconstrained linkages (Harrison et al., 2011)
point, and we examine how generative making rules for both constructing and modifying physical models can be prescribed. These rules are then systematically varied, to explore a wider space containing design variations with similar motions to the original design. Here we report on an episode of exploratory making where we implement this method, selecting a mechanical toy that exhibits an unusual type of motion, as a starting point for exploration of related kinematic designs. We consider how this artefact’s shape, structure and motion can be described, and also how making rules, which describe actions for combining and transforming materials to recreate these properties, can construct physical copies. We then vary these making rules to create new designs, and examine their motions through manipulating physical models. We also examine how further alterations prompted by the making and exploration process can be applied directly to the models. We find that manipulation of physical models not only increases an appreciation of motions, but is also an integral part of the making process itself. For the case considered here, a process of systematic experimentation using physical models yields a set of making rules, which in addition to describing actions to transform and join materials, also describe the manipulations required during making. A making rule can then be generalised into a schema, defining a wider set of rules of a particular type. A specific subset of the set of making schemas generates a set of designs that each exhibit similar motions to the original object. Once identified through making and testing physical models, this subset can also help to explain which relationships between shape and structure are essential for preserving motions for this class of designs.
Exploratory making: Shape, structure and motion
53
1
Actions and reflections, plans and tactics
Different combinations of tools and materials afford opportunities in design for specific types of things to be made and distinct types of properties to be examined. Making supports the development of ideas through constructing and modifying tangible representations. It involves reasoning about relationships and connections between materials and shapes, whilst considering the effects of design changes on properties of interest. Experienced designers may rely on tacit knowledge and established techniques to help recognize opportunities for design changes and anticipate their effects, in order to select useful or meaningful actions (Lawson, 2004). However, in a new situation where previous experience is lacking, making and testing possible changes to designs quickly helps to build knowledge about the effects of these changes on various properties. Tool and material combinations may also be altered and modified as exploration progresses, to make new things visible. For three-dimensional designs, spatial interaction is necessary to examine physical objects in the round, and tactile examination may also be important. In contrast to automated processes such as digital fabrication, exploratory making affords physical action, sensory observation, and feedback. It presents opportunities to notice and pursue new ideas and directions not conceived at the outset. These characteristics are not unique to three-dimensional physical making and share features with sketching. Sch€ on (1992) studies how 2D sketching activities inform design reasoning, and describes design as a process of action and reflection, or a ‘conversation with the materials of the design situation’ (1992, p. 3). Actions make changes to tangible descriptions of designs. Reflection involves visual examination to recognize shapes and consider their formal and semantic properties. Reflecting on the desirability of properties as they emerge both identifies and motivates opportunities for further actions or changes (Sch€ on, 1992; Sch€ on & Wiggins, 1992). In other modes of making, observations to support reflection may require more that just visual examination. Knight and Stiny propose that making activities and associated reflection potentially involve a full spectrum of senses (Knight & Stiny, 2015). Actions and reflections may occur simultaneously, making them difficult to identify separately when observing designers at work. Studies of situated activities in other domains may also be relevant when considering design exploration. Suchman (1987) explains how plans (abstract, verbal descriptions of activities) may be used to predict and, to some extent, structure interactions in real-world situations before they occur, or as a means to report on them after they have occurred. De Certeau (1988), in describing activities and interactions occurring in everyday life, distinguishes between strategies, which are implemented in a top-down manner by those who have ownership of a given space, and tactics, which provide a contingent, time-
54
Design Studies Vol 41 Part A November 2015
based approach to operating in a space that is not owned, understood or predictable. Here we consider how these concepts can be applied to the spaces of material interaction where making and design reasoning occur. In this case, we interpret ownership of a design space to imply a familiar design situation, where the effects of actions on properties of interest can be readily anticipated. In such a situation, Suchman’s plans and de Certeau’s strategies can be aligned, and generative schemas can be deployed in an unsupervised manner to create new design variations. De Certeau’s tactics are employed when continually responding and reacting to the materials of the situation, to notice opportunities for action, exploit them, and observe the consequences. In this sense, they appear similar to Sch€ on’s actions and reflections, with one key distinction: although both strategies and tactics may operate simultaneously in a given physical space, they afford a binary distinction. For de Certeau, these are the separate modes of operation of distinct groups from different political spheres, which have little scope for translation between them. In contrast, in design exploration there exists a continuous spectrum between the new and unpredictable and the familiar and understood, where the continued application of tactics may eventually afford the development of strategies. Repeated actions and reflections support situated learning about the nature of a certain situation, developing a growing appreciation of inherent relationships between properties, and improving confidence in the reliability of a particular approach. But even when reliable sequences of actions become practised and polished by the rhythms of repeated application, how can they be described, documented, or discussed? Communicating verbally the detailed nuances of material making, even when the maker has developed a practised intimacy with the activity, and all its opportunities for error, is often not straightforward. Making seems best demonstrated through practice, in a language of tools and materials, which together afford precise motions, operations, and sensations. When described out of context, all complex situated activities must become to some extent simplified into abstract schemes, lacking the essentially unbounded contextual information that defines how activities unfold in practice. In Suchman’s outlook, plans describe the steps or stages of an activity to the extent that they can be meaningfully articulated, either in advance or as a means of reporting after the fact. Situated actions then develop a finer resolution as appropriate, in direct response to a situation. For human reasoning this requirement for finer resolution in description is no obstacle to the fluent conception and implementation of appropriate responses as circumstances present themselves. This presents a key conceptual distinction between situated actions and exploratory tactics: practised situated actions, although
Exploratory making: Shape, structure and motion
55
they cannot be articulated easily in full detail, are routinely relied upon to deliver successful results, whereas with exploratory tactics, outcomes are not necessarily predictable. This study focuses on describing and documenting moves and actions by which designers explore and develop understanding in unfamiliar territories of making. We use the example of kinematic designs, where technical and conceptual difficulties in predicting outcomes call for a more effective, exploratory making approach.
2
Generative design exploration
Generative design descriptions use rules to define sets of transformations. When applied to tangible media or materials, generative rules define ‘instruction sets’ for processes to synthesise tangible things. In exploratory making, synthesising tangible designs, that is, physical artefacts, and then observing and evaluating their properties directly, provides a measure of separation between action and reflection, whilst retaining continuity in the process of design development. Generative descriptions alone may not afford an appreciation of the properties of the designs they describe, but they can make it easier to make observations about the transformation processes that construct them, facilitating discussions about design processes.
2.1
Shape rules and schemas
As shape descriptions of designs can, by definition (Stiny, 1981), allow division into parts in infinitely many ways, activities such as sketching that support the development of unstructured shape descriptions allow designs to be continually reinterpreted. As shapes are added or selected and transformed, they may interact or merge with other shapes, in ways not predictable in advance. Shape rules can be used to describe and document the distinct moves by which shapes within designs are recognised and transformed: rules identify spatial relationships between shapes and apply transformations to shapes and subshapes within designs (Stiny, 2006). Shape grammars employ sets of shape rules to construct complete designs, and can describe sets of designs that exhibit distinct visual styles or properties (Flemming, 1987; Koning & Eizenberg, 1981; Prats, Earl, Garner, & Jowers, 2006; Stiny & Mitchell, 1978). Shape schemas provide general descriptions of classes of rules, which can be instantiated in many ways, to create specific rules for particular applications. Stiny (2011) identifies several levels of general schemas, classifying types of shape rules, including for example parametric variations, and geometric transformations such as rotations, reflections, and translations. Schemas can also be specified for joining and dividing parts, selecting shape boundaries, or changing dimensions of representation by creating shapes from their boundaries and vice versa. The identity schema provides a special case, because it
56
Design Studies Vol 41 Part A November 2015
recognises shapes but does not apply further transformations (Stiny, 2011, 2006). These ‘useless’ rules at first glance do not appear to do anything, because they do not visibly transform designs. However, for structuring shape descriptions, these rules permit new shapes to be noticed or identified (Stiny, 1996). Shape rules have been used widely for the analysis of existing designs (e.g. Flemming, 1987; Koning & Eizenberg, 1981; Stiny & Mitchell, 1978). By considering the spatial relationships encountered within these designs, underlying rules can be inferred. Sets of such rules define grammars that can be used both to reconstruct existing designs and to discover new variants. Knight and Stiny (2001) also advocate an alternative, more open-ended use. Shape rules support systematic yet creative exploration and can be employed in an open-ended way (Knight, 1992, 1995, 2005; Stiny, 1980).
2.2
A method for exploratory making
We outline a method for using generative rules in a design exploration process, for the example of kinematic designs (that is, shapes with moving parts). In contrast to employing sequential actions and reflections to develop a design along a single trajectory, we consider how systematic exploration, from an existing design as a starting point, can uncover other related designs that exhibit similar characteristics e in our example, similar motions. Through experimentally varying shape and structure in physical models, our aim is to discover which relationships in an original design are essential to preserving motions, and conversely which design variables might be freely varied as required by a functional application. First, we define a set of making rules, which describe the actions and manipulations required to fabricate a working copy of an existing mechanical artefact. Systematically varying these making rules, through abstracting them into a more general set of making schemas, yields a set of physical models that embody a wide range of variations upon the original design. Evaluating the properties of each, through manipulating and observing motions, develops a more precise understanding of the scope for transformations within design constraints on kinematic behaviour. Identifying particular subsets of making schemas that reliably reproduce the motions of the original artefact helps to explain how a particular design works.
2.2.1
Selecting a starting point
Our approach employs an existing kinematic design as a starting point. Designs appropriate for the method are those that are not already well-known and well-explored by conventional analysis techniques. Kinematic designs with distinctive, unusual motions are also advantageous because the presence or absence of distinctive motion characteristics will be apparent immediately. To avoid explorations becoming limited artificially by expectations about an
Exploratory making: Shape, structure and motion
57
object’s function, designs that have no familiar functional applications are preferred.
2.2.2
Manipulating physical models
Physical models of designs with moving parts afford an interrogation of possible motions through manipulation. Certain variable spatial relationships are inherently afforded by particular designs, but these must be elicited and observed through direct interaction, so that the relative success of a modified design in reproducing these motions can be evaluated. In some cases modified kinematic designs exhibit restricted motions (perhaps through transformations of shape causing collisions, or transformations of connections causing reduced mobility). Descriptions of the configuration states into which the physical model of the kinematic design can be manipulated provide a means to describe motion characteristics.
2.2.3
Describing a design in terms of shape and structure
The shapes of parts, and the types of the connections between them in a kinematic design, describe the shape and structure of the design. Shape rules afford an immediate route to describing shapes and connections (Stiny, 1981).
2.2.4
Making rules for motions
Certain modes of material combinations allow for the fabrication of specific connection types more readily than others. For hinged designs, for example, simple materials readily afford the construction of physical models that exhibit appropriate variable spatial relationships between parts. More sophisticated connection types may require more advanced tools, materials, and fabrication processes. Here, making rules are used to describe the actions and manipulations employed for shaping and connecting materials whilst constructing physical models.
2.2.5
Abstracting making rules into tactical schema
Removing detail from making rules for shape and structure can define making schemas, describing sets of rule variations. Particular combinations of these new rule variations can then be employed to construct physical models of new designs that are related to the original object.
2.2.6
Evaluating motions of new designs
New models can then be manipulated using the actions discovered when interacting with the original artefact. This enables evaluation of the extent to which motions are preserved in the new design.
2.2.7
Situated tactics
Whilst the original making rules provide a starting point from which to derive new rule variations and construct new designs, new physical models
58
Design Studies Vol 41 Part A November 2015
themselves may prompt additional ideas for alterations that can be applied directly to the objects.
2.2.8
Discovering strategic schemas
Through systematically constructing and considering model variations, subsets of the tactical making schema (see Section 2.2.5) that create new objects which reliably reproduce the motions of the original design are identified.
3
Rules for making
Whilst shape rules are useful for describing aspects of designs with moving parts, they are not sufficient to fully encode the types of action and reflection needed to construct and evaluate them. Knight and Stiny (2015) suggest that the concept of identity rules for recognizing shapes could be expanded to create more general ‘sensing’ rules to describe other forms of observation and reflection, and also that shape rules can provide a basis for defining different types of ‘doing’ rules for transforming materials. Sets of doing and sensing rules can be employed within making grammars, to describe making activities of particular types. Here, we consider how shape rules can be adapted to define making rules for kinematic designs. As doing and sensing can occur simultaneously, making rules must support both action and reflection. We identify two specific types of rule, which are essential for considering kinematic designs. One of these defines the creation of various types of connections between parts, which possess various classes of variable spatial relationships. The other type describes manipulative actions and interactions on models. These manipulation rules determine the overall motions afforded by the combined effects of the variable spatial relationships defined by each connection.
3.1
Describing structure and variable spatial relationships
For three-dimensional models composed of only rigid materials, joints or connections between pairs of rigid parts impose constraints on their relative motions. These connections are termed kinematic pairs. In lower kinematic pairs (Figure 2), connected parts share surfaces of contact, whereas higher pairs exhibit point or line contact (Reuleaux, 1876). Shape rules might be extended to describe subshape relations that afford variable spatial relationships between parts in kinematic pairs (Figure 4). For lower kinematic pairs, subshape relations can use surface embedding to define parts of surfaces as shared sub-shapes within both parts. Motions observed in lower kinematic pairs due to these shared surfaces can also be modelled by shape representations using sub-shapes of lower dimensions, such as shared lines and points (see Figure 3).
Exploratory making: Shape, structure and motion
59
Figure 2 Lower kinematic pairs
Figure 3 Connections that create variable spatial relationships described by shared sub-shapes
60
Design Studies Vol 41 Part A November 2015
Figure 4 A shape rule describing a variable spatial relationship
3.2
Manipulation
The potential for parts to move within a design is directly related to the composition of its connections and the shapes of its parts. Descriptions of shape and structure therefore inherently contain information about a design’s motions. However, when many variable spatial relations interact, the overall motions that result may not be straightforward to determine. Physical models, embodying three-dimensional descriptions of both shape and structure, provide an expedient way to both examine and describe tangibly the motions occurring in these designs (Video S1a and b). Interaction with unfamiliar physical objects is both tactical and situated, as manipulation itself is exploratory. However, through a process of experimentally applying forces, particular sequences of actions that successfully
Video S1a.
Exploratory making: Shape, structure and motion
61
Video S1b.
articulate a design’s full range of motions may come to be understood, and these can then be systematically applied. In further design exploration, practised manipulation sequences may be experimentally applied to new designs, to see if motion sequences encountered previously are preserved. Because manipulation rules are situated and act directly upon models, it is difficult to define them out of context. Further, as the motions they elicit from models may be complex, it may not always be possible to succinctly describe manipulations in terms of the variable spatial relations they enact. Here, rather than attempting to describe the manipulation rules required for design exploration in full detail, we consider that, from a set of all possible manipulations, a particular sub-set is discovered to successfully elicit the motions of a given artefact.
4
An example of exploratory making
We report on an episode of systematic exploration through making, implementing the method outlined in Section 2.2. This practical work was undertaken by the first author. The actions, interactions and reflections that occurred were observed and documented at every stage. In addition, as care was taken to construct a new model each time a design change or alteration was applied, the resulting collection of models provides a lasting physical record of the design activity that was undertaken.
4.1
Selecting an existing object as a starting point for design exploration
The chosen starting point was a mechanical toy that exhibits an unusual sequence of motions, which fulfil no immediate function beyond their ability
62
Design Studies Vol 41 Part A November 2015
Video S2.
to entertain. The toy is composed of two connected layers of flat, hinged panels that form a rectangular sheet (see Video S2). We encountered this object as a sample model from a Selective Laser Sintering equipment manufacturer. A 3D printable copy is available (Thingiverse, 2013). We have not yet discovered the designer of the original. However, we discovered a patent for a similar design that exhibits an identical sequence of motions, but is composed of only one layer, and intended to be folded from a single sheet of card (Byrnes, 2004). We expect that the two-layer version may be an intentional variation, in order to achieve the same motion sequence as the protected design without infringing on the shape and structural descriptions specified in the patent.
4.2
Discovering and describing motions through manipulation
The toy’s hinged panels can be manipulated into four distinct states. When manipulated continuously, the toy transforms between these states consecutively in a continuous cycle. Figure 5 illustrates the sequence of unfolding actions through which the toy eventually returns to its initial state. This type of motion sequence is referred to as full-cycle mobility (Byrnes, 2008). This toy design exhibits non-continuous motions, as symmetric pairs of hinged panels can fold through a maximum of 180 before shape interference limits motion. (In contrast, the designs in Figure 1 and Video S1a and S1b exhibit continuous full-cycle mobility.) During each unfolding action, two symmetric hinged segments are rotated until a limit to motion is reached. From this new state, new sets of hinges become aligned, and a new phase of motion becomes possible.
Exploratory making: Shape, structure and motion
63
Figure 5 Motion sequence between the four states (note that sequence may also occur in reverse)
The effects of the manipulation sequence upon shape and structure can be described by examining the toy in its four possible states from different perspectives, and then creating drawings as shown in Figure 6. This non-continuous behaviour is relatively uncommon. The nature of these motions was unanticipated upon first encounter with the object, but became apparent to us, and to several others informally presented with the artefact, after some initial exploratory manipulation. Practice then rendered familiar the sequence of actions required to cycle through states continuously. Within the following exploration exercise, this sequence of actions was used to test whether the new physical models created also exhibit similar motions.
4.3
Describing spatial relationships between shape and structure
Whilst considering all reachable states or configurations helped to describe and examine motion, the toy’s shape and structure (that is, its parts and connections) alone could be represented just by the appearance of the toy in State 1. In this state, the toy resembles a flat sheet (Figure 7). Drawings of each face in this configuration were relatively simple to produce, and embody a large number of particular spatial relationships (see Table 1, column A).
64
Design Studies Vol 41 Part A November 2015
Figure 6 Drawings of the toy design in its four distinct states
4.4
Rules for making
From this description of shape and structure, we developed making rules for each layer. To construct a working copy of the design, common modelmaking materials that permitted rapid construction and modification of new models were employed. Because the design is composed of flat, hinged panels,
Figure 7 Spatial relationships in the original design
Exploratory making: Shape, structure and motion
65
Table 1 Spatial relationships in (A) the original design, (B) as simplified after testing the effects of rule variations on motions, and (C) the essential relationships discovered through exploration (in bold) with other related descriptions that support systematic making (in parentheses)
A. Initial spatial relationships
B. Related lesser statements
C. Essential relationships (and related lesser descriptions to support making)
The overall shape of the design is a The overall shape is a rectangle. rectangular sheet, twice as long as wide. The edges of the overall shape are straight.
(The overall shape may be any closed curve that contains the hinge lines and their intersections.)
The hinge-lines on a given face are parallel to the edges of the sheet, and therefore also to each other.
Hinge-lines within pairs on each face must be parallel to each other.
Hinge-lines within pairs on each face must be parallel to each other.
The hinge-lines are positioned 1/4 and Pairs of hinge-lines are positioned (Interfering material may need to 3/4 along the face. symmetrically about a centre-line. be tactically removed.) The hinge-lines are spaced apart by half the panel’s full width/length. Hinge pair orientation and spacing is identical but perpendicular on each face.
Pairs of hinge-lines must be Pairs of hinge-lines must be perpendicular to those on the opposing perpendicular to those on the face. opposing face.
The split-line must be perpendicular to The split-line is a straight line. the hinge-lines. The split line or curve on one face is perpendicular to the split line on the opposing face. The design is composed of 2 full sheets of material.
(The split-line or curve on a given face must bisect both hinge lines on that face.) (Enough material must be present to maintain the object’s integrity and support attaching the two layers together.)
we used foam-board to cut 2D shapes, and tape to selectively re-join panels along cut-lines to instantiate hinges. Shape rules, with conventions for describing hinge-lines, could then describe how to shape and join materials (Figure 8). Whilst these do not describe the full details of the complex practical activities of cutting and joining, this contextual knowledge could be accessed readily in the situated application of schematic rules for making. External examination alone, however, did not identify how the two panel layers, assembled using Rules 1, 2 and 3, should be joined together. Situated experimentation yielded this information. Manipulating the two assembled sections established which parts must move and therefore should not be attached. This manipulation was conducted in parallel with the tentative application of adhesive at various points to test whether the model would still transform as expected. Figure 9 demonstrates schematically a reasoning process that determines a rule for where adhesive can be applied. We observed that tactics of trying and testing alone discovered the same result. Figure 10 describes this additional making rule, required for the complete assembly of a working
66
Design Studies Vol 41 Part A November 2015
Figure 8 Rules for making
Figure 9 A reasoning schematic to derive Rule 4 (for joining the two panel layers)
model. Manipulation through the four states confirmed that this model reproduces the motions of the original toy.
4.5
Varying making rules to give tactical schema for design exploration
More models were constructed by varying the original rules to explore and test which aspects of the original design are important for preserving motions. Note that at the outset we had no intuition as to which variations would be successful, but we attempted to explore a wide range of possible changes systematically. We considered how rules might be simplified and abstracted into more general schemas, from which new variants on the original making rules could be defined. For each variation, we constructed a new physical model. Figure 13 shows the full set of models constructed (Models 1e14). The first variation (Model 1) generalized the proportions for the rectangle in Rule 1 to allow parametric variations. When the overall shape is set to a square,
Exploratory making: Shape, structure and motion
67
Figure 10 A rule 4 for making: gluing layers
both faces become identical, and the schematics for Rules 2 and 3 become simplified (Figure 11). In this case, for assembly, one layer must be rotated through 90 relative to the other. Through allowing parametric variation, Rule 10 evolves into Schema 10 (Figure 12), which constructs both square and rectangular panels of all proportions (for example Model 14). Schema 100 is further abstracted, to allow for non-rectangular shapes (Figure 12; Figure 19). A second variation tested whether the pairs of hinge-lines created by Rule 2 must be located as precisely as originally specified, or whether a more general schema (Schema 2, Figure 12) would be acceptable. In the original design, outer edges meet precisely along a centre-line upon folding inwards. Avoiding
Figure 11 From Rule 1 to Schema 1 and simplified rules 2 and 3
68
Design Studies Vol 41 Part A November 2015
Figure 12 From making rules to making schema
overlapping and associated interference requires that the spacing between the hinge lines must be at least half the width of the face, but a variation that relaxed this rule so that hinges might also be further apart, was also tested (Models 7 & 8). Further variations tested whether hinge-lines must be symmetrically placed, deriving Schema 20 (Figure 12; Models 13 and 14), whether hinge lines must lie parallel to the panel edges (Schema 200 , Figure 14; Model 4), or parallel to each other (Model 12). When varying hinge alignment, external panel edges can meet at an altered centre-line upon folding, and the greatest possible deviation would orient this centre-line along the diagonal of 45 . To achieve this, hinge lines were rotated by half this angle (Models 4, 5, 10, 11). To preserve the perpendicularity of hinge alignment between the faces during this variation, the layout of one face must be a mirror image of the other, rotated by 90 , rather than an absolute copy (Model 9 tested whether orientation between the 2 layers could also be varied). Further variations (Figure 12) tested how Rule 30 might be simplified: by removing the requirement that the split-line be perpendicular to the hingelines on that face (Models 4 and onwards; Schema 3); removing the requirement that it must be a straight line (Schema 30 ; Models 13 and 14); and also testing whether split-lines must be placed similarly on both faces (Models 13 and 14).
4.6
Evaluating motions of new models through manipulation
After fabrication, each new model was manipulated using the same sequence of actions as was discovered to elicit the motions of the original toy (see Section 4.2). When a new model could be manipulated successfully through all four states sequentially, the rule variations that constructed it were deemed to be allowable for preserving those spatial relationships essential for achieving the motions exhibited by the original design. Table 1 outlines how
Exploratory making: Shape, structure and motion
69
Figure 13 Experimental models to test design variations
70
Design Studies Vol 41 Part A November 2015
Figure 14 Schema 5 illustrates a tactic (consequent on Rule 200 ) to remove and reattach interfering material. The notation M denotes a set of manipulations
this new knowledge of successful variations allows the descriptions of spatial relationships observed in the original design to be relaxed, creating less precise descriptions that define wider categories of objects, and ultimately identifying essential relationships for achieving motions.
4.7
Applying situated tactics to directly modify objects
In Section 4.4, we noted the need for situated, tactical experimentation in order to derive Rule 4. This need for a situated reaction to models, to conceive of new making rules, continued to occur throughout our process of exploration. This was particularly important when rule variations initially appeared to create unsuccessful models, which could not be manipulated through the full four-state cycle. When Schema 200 was applied (see Figure 14) the resulting model (Model 4a) could be manipulated into states 2 and 4, but material interference prevented state 3 being reached. However, here a new idea for a situated rule was identified: when examining the design in states 2 and 4, its footprint was noticed to be larger than its square shape boundary in State 1. A tactic of trimming sections that protruded beyond the original square footprint was tried. This resulted in a model that could then reach all four states successfully (Model 4b).
Exploratory making: Shape, structure and motion
71
It was also noticed later that this procedure could be further refined, and slightly less material could be removed, while still yielding a working design (Model 10). Interfering material could be marked precisely, directly on the model, in states 2 and 4 respectively, and these marked sections could then be removed. It was found latterly that this trimming approach could be applied successfully to complete any design variation where material interference was initially encountered. Because this occurred for a significant subset of the designs produced here, this new schema, Schema 5 (Figure 14), became an important addition to the making sequence. Although conceived through situated experimentation, once practised and found to be a reliable method, this schema could be defined precisely and applied systematically. Note that the original making rules assemble the design exclusively in State 1. However, to discover and remove interfering material, Schema 5 relies on subsequent manipulations, using the set of manipulations M to transform the model into and between states 2 and 4 (see Figures 5 & 6), so that marking and trimming operations can be applied. Therefore actions for manipulating designs between states (see Section 4.2) are required not only for testing the success of designs, but also as part of the making process itself. Further situated tactics were uncovered by considering how successful models might be altered further, by selectively removing material. Rule 4 (Figure 10) indicated that material that lies outside of the hinge-lines plays a role in connectivity between the two layers, but material lying inside the layers might be less essential. Through experimentally removing this material, we ascertained that this was indeed the case (Model 2) and defined a corresponding Schema 6 (Figure 15). Further, considering the resulting model in its intermediate states indicated that further portions of material also played no role in essential connectivity. Also removing these parts led to a pared-down model (Model 3), which changes its footprint dramatically upon transforming between states (Figure 16). These reducing rules were found to also be applicable to other model variants, producing a subset of designs that possess the additional property of expanding and collapsing between states (Models 5, 7, 8). These pareddown designs could also be constructed from a single sheet of thin material. They therefore appear to fall within the subset of designs protected by the patent. Note that the resulting designs do not require full sheets of material in their construction. Thus, they could also be constructed more efficiently from their necessary parts directly using Schema 7 (Figure 17).
4.8
Strategic schemas for making
After making a number of models, we established that many of the relationships inherent in the original design were inessential for preserving its motions.
72
Design Studies Vol 41 Part A November 2015
Figure 15 Schema 6 e removing material
Figure 16 The two pared-down models (3 and 5), in states 1e4 with expanding and collapsing footprint
Figure 17 Schema 7 to assemble wholes from parts
Exploratory making: Shape, structure and motion
73
Significantly, the only relationships that proved essential were the relative orientations of the hinge-lines, which define a particular set of variable spatial relationships between the design’s parts (see Table 1, column C: bold text). Models 9 and 12 (Figure 13), in which hinge orientations were altered, were found to be unsuccessful, not due to material interference, which can be fixed using Schema 5 (Figure 14), but due to structural misalignment. This apparent importance of hinge orientation indicated that Schema 2 (Figures 12 and 14), which defines hinge geometries, is perhaps the most critical step in the making process. Because other design aspects are constrained only by their locations relative to these hinge lines, placing Schema 2 at the beginning of the making sequence defines a more strategic approach to constructing designs. An evolved Schema 2* (see Figure 18) initially defines relations between hinges for both layers. As the remaining panel edges may otherwise form any closed shape, a corresponding Schema 1* must merely place material outside the hinges and their intersections to ensure practical construction. A Schema 3* adds a split-line or curve to each face, bisecting both hinges. As before, Schema 4 is required for gluing, and Schema 5 is sometimes necessary for removing material interference. These new, strategic schemas, outlined pictorially in Figure 18, permit a more direct, strategic approach to constructing further designs. With the allowable limits to variations of shape and structure now clearly defined, creative exploration can focus on other design properties, such as colour and composition (Figure 19). Formally defining the manipulations and alterations that occur within Schema 5 can also support further creative exploration using digital drawing tools. The interplay between exploration in physical making and in other representations, including digital tools, is the subject of further work by the authors.
5
Discussion
This exercise in exploratory making goes some way towards demonstrating a possible role of a generative approach in developing understanding of relationships between shape, structure, and motion in kinematic designs. For the case of the single design considered here, practical design reasoning using exploratory model-making successfully derived sets of making schemas that constructed larger sets of designs with similar properties. Further, these schemas also possess an explanatory function, illuminating how or why the original design works. Significantly, models that embodied unsuccessful changes were often more useful for positively identifying essential relationships than successful models.
74
Design Studies Vol 41 Part A November 2015
Figure 18 Schemas for strategic making
Usually seen as abortive experiments, these unsuccessful results of design exploration played a critical role in both identifying and explaining essential relationships. In our study, alterations that affected spatial relationships between hinge-lines were discovered to exceed allowable limits for variations. This discovery assisted the positive recognition of boundaries within which motions are preserved, thereby allowing reliable generative schemas to be developed. These schemas, in turn, helped to explain which aspects of shape and structure found in the original design were important for preserving its motions in new design variations. The design we began with possessed unusual and distinctive motions, and therefore the presence or absence of these full-cycle sequences in new design models could be evaluated readily. A single example has been considered here, but further work to consider other classes of designs could establish the extent to which this approach might be applied more widely. We expect that the method is transferable most easily to types of kinematic design composed of closed linkages, where the overall motion of the design is highly constrained. In our example, unsuccessful designs were those where full-cycle motion sequences were no longer possible. The binary nature of this test makes it far easier to apply than detecting subtler changes in more complex, or lessconstrained, kinematic behaviours. Techniques and language for describing detailed motions in designs directly are not commonly available, and description in this paper focused primarily on shape and structure. We note that manipulation of physical models to
Exploratory making: Shape, structure and motion
75
Figure 19 Creative designs developed using the evolved making schemas
examine motions was at first tactical and exploratory in nature, but later a practised sequence of manipulations became familiar and routine. This example highlights the importance of physical manipulation when considering kinematic designs with moving parts. Manipulation develops an understanding of motions, tests the success of design modifications, and also plays a role within the synthetic steps of the making process. Physical manipulations and associated exploratory making are a situated set of actions and evaluations. Situated behaviours in other fields have helped to provide a framework for considering the tactical manner in which appropriate manipulation sequences are discovered through material interaction.
76
Design Studies Vol 41 Part A November 2015
Because anticipating effects of changes to shape and structure is difficult in the kinematic design we considered, the tactical nature of exploration assumes a particularly important role. The importance of tangible models, for affording opportunities and ideas for new transformations and modifications, also became apparent, as these new directions could not be readily conceived through considering abstract descriptions alone. Further, our collection of experimental physical models also seems to play a useful role in presenting this work to others. As tangible representations of motions, they provide an expedient route for disseminating the explanations about relationships between shape, structure and motion that were uncovered by our generative approach.
References Boden, M. A. (2004). The Creative Mind: Myths and Mechanisms. London: Routledge. Byrnes, R. B. (2004). Playthings. U.S. Patent No. 20,040,092,199. Washington, DC: U.S. Patent and Trademark Office. Byrnes, R. (2008). Metamorphs: Transforming Mathematical Surprises. Tarquin Publications. de Certeau, M. (1988). L’invention du quotidien. In: Arts de faire, Vol. 1. Berkeley, CA: University of California Press. Flemming, U. (1987). More than the sum of parts: the grammar of Queen Anne houses. Environment and Planning B: Planning and Design, 14(3), 323e350. Harrison, L. C. A., Earl, C. F., & Eckert, C. M. (2011). The role of rapid prototyping in the generative design of mechanisms. In 12th Conference on Rapid Design, Prototyping & Manufacturing (RDPM 2011). Hunt, K. H. (1978). Kinematic Geometry of Mechanisms. London: Oxford University Press. Ingold, T. (2009). The textility of making. Cambridge Journal of Economics, 34(1), 91e102. Ingold, T. (2013). Making: Anthropology, Archaeology, Art and Architecture. London: Routledge. Knight, T. W. (1992). Designing with grammars. CAAD Futures’ 91: Computer Aided Architectural Design Futures: Education, Research, Applications 33e48. Knight, T. W. (1995). Transformations in Design: A Formal Approach to Stylistic Change and Innovation in the Visual Arts. New York, NY: Cambridge University Press. Knight, T. (2005). Creativity rules. In J. S. Gero, & M. L. Maher (Eds.), Computational and Cognitive Models of Creative Design VI (pp. 155e174). University of Sydney: Key Centre of Design Computing and Cognition. Knight, T., & Stiny, G. (2001). Classical and non-classical computation. Arq: Architectural Research Quarterly, 5(04), 355e372. Knight, T., & Stiny, G. (2015). Making grammars: from computing with shapes to computing with things. Design Studies, 41(PA), 8e28. Koning, H., & Eizenberg, J. (1981). The language of the prairie: Frank Lloyd Wright’s prairie houses. Environment and Planning B, 8(3s), 295e323. Lawson, B. (2004). Schemata, gambits and precedent: some factors in design expertise. Design Studies, 25(5), 443e457.
Exploratory making: Shape, structure and motion
77
O’Rourke, J. (2011). How to Fold It: The Mathematics of Linkages, Origami, and Polyhedra. Cambridge University Press. Phillips, J. (2007). Freedom in Machinery. New York, NY: Cambridge University Press. Prats, M., Earl, C., Garner, S., & Jowers, I. (2006). Shape exploration of designs in a style: toward generation of product designs. AIEDAM, 20(03), 201e215. Reuleaux, F. (1876). The Kinematics of Machinery: Outlines of a Theory of Machines. London: Macmillan & Co. Sch€ on, D. A. (1992). Designing as reflective conversation with the materials of a design situation. Knowledge-Based Systems, 5(1), 3e14. Sch€ on, D. A., & Wiggins, G. (1992). Kinds of seeing and their functions in designing. Design Studies, 13(2), 135e156. Stiny, G. (1980). Kindergarten grammars: designing with Froebel’s building gifts. Environment and Planning B, 7(4), 409e462. Stiny, G. (1981). A note on the description of designs. Environment and Planning B, 8(3), 257e267. Stiny, G. (1996). Useless rules. Environment and Planning B: Planning and Design, 23(2), 235e237. Stiny, G. (2006). Shape: Talking About Seeing and Doing. Cambridge, MA: The MIT Press. Stiny, G. (2011). What rule (s) should I use? Nexus Network Journal, 13(1), 15e47. Stiny, G., & Mitchell, W. J. (1978). The Palladian grammar. Environment and Planning B, 5(1), 5e18. Suchman, L. A. (1987). Plans and Situated Actions: The Problem of HumanMachine Communication. New York, NY: Cambridge University Press. Thingiverse. (2013). Flipping Folding Fun Toy. Retrieved from. http://www.thingiverse.com/thing:53658. Wiggins, G. A. (2001). Towards a more precise characterisation of creativity in AI. Case Based Reasoning: Papers from the Workshop Programme at ICCBR, 1, 113e120.
78
Design Studies Vol 41 Part A November 2015