- Email: [email protected]
Robotic systems for Nasa ground-based research
19
Autonomous Systems for Space
Robotic Systems for NASA Ground-Based Research Joseph N. H e r n d o n and William R. Hamel Oak Ridge National Laboratory *, Oak Ridge, Tennessee 37831, USA
and
Alfred J. Meintel NASA Langley Reseaech Center, Hampton, Virginia 23665, USA The realization of the National Aeronautics and Space Administration (~ASA) Space Station Program will mark the beginning of a major new era in space exploration and utilization. This new era will be identified by much increased utilization of dexterous robotics systems to reduce the need for hazardous extravehicular activities by astronauts. The goals are to improve overall system safety and efficiency as well as provide significant spin-off technology to improve the productivity of the US industrial sector. In support of these goals, the Laboratory Telerobotic Manipulator (LT~) is being developed at ORm. under the direction of IqAS^ Langley Research Center to provide telerobotic hardware for use in NASA ground-based laboratory research. A significant technical limitation is the lack of available telerobotic hardware that can function well as a real-time teleoperator and can also provide a sound hardware basis for intelligent robotic operations. The LTM is being developed to merge these technical domains in common hardware to further NASA'S research investigations. This article summarizes the mechanicul and controls approach being utilized to realize these goals.
1. Introduction
The realization of the National Aeronautics and Space Administration (NASA)Space Station Program will mark the beg0nning of a major new era in space exploration and utilization. This new era will be identified by much increased utilization of dexterous robotics systems to reduce the need for hazardous extravehicular activities by astronauts. The goals are to improve overall system safety and efficiency as well as provide significant spin-off technology to improve the productivity of the us inclustrial sector. In support of these goals, the Laboratory Telerobotic Manipulator (LTM) is being developed at Oak Ridge National Laboratory (ORNL) under the direction of NASALangley Research Center (LaRC)
Joseph N. Herndon is presently Manager of the Laboratory Telerobotic Manipulator Program at Oak Ridge National Laboratory. Mr. Hemdon has over 10 years experience in the fields of remote handling system design and operation, remote facility design, and manipulator research and development. Mr. Herndon has BS and MS degrees in Mechanical
Engineering.
Keywords: Telerobot, Force reflection, Modular, NASA, Traction drives.
William R. ~ is presently Section Head of the Telerobotic Systems Section at Oak Ridge National Laboratory and is a principal investigator in the op.m. Center for Engineering Systems Advanced Research. Dr. Hamel has over 15 years experience in controis and electromechanical systems development. Dr. Hamel has BS, MS and PhD degrees in Mechanical Engineering.
* Research performed at Oak Ridge National Laboratory, operated by Martin Marietta Energy Systems, Inc., for the us. Department of Energy under Contract No. DE-AC05840R21400, and sponsored by the National Aeronautics and Space Administration. North-Holland Robotics 4 (1988) 19-25
Alfred J. Meintei is presently Head of the Automation Technology Branch within the Information Systenas Division at ~ A S A Langley Research Center.
0167-8493/88/$3.50 © 1988, Elsevier Science Publishers B.V. (North-Holland)
20
J.N. Herndon et a L / Robotic Systems for Ground-Based Research
to provide telerobotic hardware for use in NASA ground-based laboratory research. Present NASA plans indicate the need to rely on teleoperation for control of these dexterous systems in the construction and early operation of the Space Station. Evolution to intelligent autonomous robotic operations will be gradual. Due to the unique nature of orbital operations, this evolution must be made in a very carefully controlled and smooth manner. A significant limiting factor in this present technical approach is the lack of available telerobotic hardware that can function well as a real-time teleoperator and can also provide a sound hardware basis for intelligent autonomous robotic operations. The LTM is being developed to form a basis for the merger of these rather diverse technical domains into common hardware to further NASA'S research progress. The LTlvlwill be a highperformance, force-reflecting servomanipulator system to maximize the approach to astronaut EVA task performance while at the same time providing the positioning accuracy capabilities and control architecture necessary for the implementation of sensor-based robotic operations as the evolutionary path toward autonomy. An additional goal is to provide a system concept that minimizes the differences between ground-based research hardware and space systems to aid in demonstrations and development. This overall technical approach is based on the conceptual design study summarized in [1].
2. Technical Approach The merger of the mechanical and control features necessary for a force-reflecting servomanipulator and a robotic positioner into a single system is a particularly difficult task. As shown in Fig. 1, a good force-reflecting servomanipulator designed for efficient human-in-the-loop control emphasizes end effector speed, high joint backdrivability for force reflection, low friction, and low reflected inertia. On the other hand, a good robotic positioner emphasizes end effector accuracy, end effector speeds, and mechanical and control stiffness. As shown in Fig. 1, the LTM design will bridge the gap between these two technologies by providing the most important design and operational parameters of each. The LTM Prototype system will comprise two force-reflecting slave
and master arm pairs with a digital-based control system providing bilateral, position-position, force-reflecting control. In addition, joint-level position and velocity robotic control and openloop joint drives will be provided for implementation of robotic control.
2.1 Mechanical Design The LTM design, as shown in Fig. 2, utilizes a modular approach for joint construction with common pitch-yaw differential joints implemented for the arm shoulder, elbow, and wrist. An output roll follows the wrist pitch-yaw differential to give a three degrees-of-freedom wrist capable of orientation within a compact hemispherical volume [2]. This construction furnishes three degrees of freedom for translational positioning capabilities and an additional redundant degree of freedom which probably will be utilized primarily for translational motion. The benefits of a redundant kinematics approach include multiple solutions for singularity avoidance and improved ability to work in physically constrained areas. This is a proper direction for NASA ground-based research. This approach does result in more possibilities for colinear axes within the arm range-of-motion and will require more complex control techniques both in the teleoperator and in the robotic mode. The LTM WriSt pitch-yaw utilizes a dual differential and traction drives for joint force transmission. The joint design is based on two input drive pinions with one output rotating about two orthogonal axes. The input pinions are driven by DC motors through the differential with variable preloading mechanism to maximize the differential efficiency. A variable preloading mechanism is also incorporated at the joint output. Each motor is equipped with a zero-backlash reducer, brake, tachometer, and optical encoder. Absolute encoders are utiliTed at the pitch-yaw joint outputs to provide high accuracy positioning for robotic operation. A torque sensor is incorporated between the reducer and the input preloading mechanisms. This torque signal will be used to provide improved joint backdrivability for bilateral, force-reflecting teleoperation. The LTM slave shoulder and elbow pitch-yaw joint modules will be similar to the wrist, but will be scaled to a higher capacity to allow for improved perfor-
J.N. Herndon et al. / Robotic Systems for Ground-Based Research
21
GOOD INDUSTRIAL ROBOT
GOOD FORCE-REFLECTING TELEOPERATOR
End effector speed 36 in./s
End effeetor speed 30 to 50 in./s
Friction 1-5% of capacity (at expense of increased backlash)
Friction 30-100% of capacity
Medium to low backlash
No backlash (at expense of increased friction)
Replica master control
Teach pendant, keyboard
I- to 2-in deflection at full load
Minimal deflection at full load (.010 to .05 in.)
6 DOF and end effector
4 to 6 DOF and end effector
Bilateral position-position control for force reflection
Force feedback with 6-axis end effector sensor
Relatively low inertia for minimum fatigue
High inertia for stiffness
Kinematics approximately manlike
Kinematics mission dependent
Accuracy and repeatability not important
Accuracy and repeatability very important
1:4 to i:i0 capacity/weight ratio
i:i0 to 1:40 capacity/weight ratio
Universal end effector
Interchangeable end effeetors
LTM End effector s p e e d ~ 36 in./s Friction close to teleeperator, much lower than robot Backlash close to robot, much lower than teleoperator Replica master control preferable, joysticks and autonomy research possible 0.030-in. deflection at full load Bilateral position-position control for force reflection Low inertia compared to robots Manlike kinematics for dexterity in teleoperation Universal interface for NASA end-effector research Capacity of 20 ib continuous, 30 ib peak Arm cross section to reach inside 6-in.-square opening
Fig. 1. L ~ design basis compared to teleoperator and robot performance.
mance in ground-based operations. The master arms of the LTM will be constructed from three wrist pitch-yaw modules since the required joint capacities can be much lower. Traction drives with preloading mechanisms were chosen for torque transmission through the LTM differentials [3]. Although traction drives have
not been widely used for servocontrol applications to date, they potentially can provide many benefits for these applications, particularly in space. One benefit is no backlash with only local creep occurring at the point of contact. Additional benefits include high torsional stiffness, good efficiencies, and reduced lubrication requirements
22
J.N. Herndon et aL / Robotic Systems for Ground-Based
~/SHOULDERPITCH 5"
Fig. 2. LTM~
~/.ELBONPITCH
layout and modularity.
compared to equivalent capacity gear drives. The penalties for traction drives include decreased volumetric efficiency and use of a technology that is less demonstrated compared to more common techniques such as gears. A full pitch-yaw test
Fig. 3. L ~ pitch-yaw joint test stand.
Research
stand unit has been fabricated at ORNL to begin development of control algorithms and to verify the traction drive design methodology. This first test stand is illustrated in Fig. 3 with the traction drive differential shown in a close-up view in Fig. 4. This test stand is presently operational at ORNL and a second test stand is nearing completion. The second test stand will allow master-slave control algorithm development in a controlled bench-top environment prior to implementation on the full LTM system. Finally, the mechanical joint concept was developed to provide a cost-effective solution for NASA laboratory research needs. A significant number of follow-on LTM systems are presently planned. The common joint design concept lends itself to reduced design and fabrication costs. Commercially available componentry is being utilized to the maximum extent practical. The entire design and fabrication effort at ORNL will utilize computer-integrated manufacturing techniques to minimize fabrication costs and to ease the NASA technology transfer process.
23
J.N. Herndon et al. / Robotic Systems for Ground-BasedResearch 2.2 Control System
Fig. 4. LrMjoint differentialand tractiondrives.
Development of the LTM control system will be a significant challenge. ORNL brings sitmificant experience in the development and implementation of digital control systems for force-reflecting servomanipulators gained through successful hardware and software development on three separate servomanipulator systems over the past six years. In addition to the control modes described earlier, the LTM control system must provide interfaces for dextrous end effectors and must provide "hooks" for alternate slave arm controllers such as universal hand controllers, joystick resolved rate controllers, optimized replica controllers, and external robotic/autonomous controllers. It is of utmost importance that the LTM control system architecture be flexible and allow layering into hierarchical control architectures such as the NASA/CNationalBureau of Standards NASREM architecture [4]. Fig. 5 illustrates the functional approach to implementing a digital-based control system for bilateral, force-reflecting control while prodding the embedded needs for implementation of intelligent robotic control techniques. The shaded blocks
Path Plenner end Sensory Transformer
Force/Torque
World
Sensory Feedback System
Fig. 5. Servomanipulator/intelligentrobot functionalcontroldiagram.
24
J.N. Herndon et al. / Robotic Systems for Ground-Based Research
FUTURELINKTO UPPERLEVELSIN HIERARCHICALSYSTEM
I..
Interface~ - ~
3
I
Fiber Optic CommunicationLink
LEFT ARM SHOULDER
RIGHTARM
~
J SHOULDER
LEFT ARM
RIGHTARM
SHOULDER L
i SHOULDER i
ELBOW WRIST
~
-! ELBOW
ELBOW
J
WRIST
~ WRIST
WRIST ROLL/ END EFF. |
vI ENDEFF.
WRIST
WRIST ROLL/ END EFF.
I
ELBOW
I
WRIST ROLL/
~Distributed OnboardElectronics( Typical )
Fig.6.LTMcontrolsystemhardwareblockdiagram. in Fig. 5 are those actually being implemented initially on the LTM Prototype system. The LTM control system hardware block diagram is shown in Fig, 6. As described earlier, the LTM will have significant numbers of signals originating within each pitch-yaw joint module. In order to reduce the arm onboard cabling bundle to a manageable size, all onboard sensor data will be locally collected within each joint module and processed onto a serial link for communication to the central control rack. The onboard data collection and communication hardware will be of custom design for this application. As shown in Fig. 6, each of the master and slave arm pairs will be provided with a central control rack. Master-to-slave and slave-to-master communication will also be handled by high-speed serial communication (10 megabits per second) over a fiber optic link. The central control racks for the master and slave will each be based on the vMEbus backplane for an open architecture on an industry standard (IEEE P1014) bus structure with rugged construction. Also, much vendor and product support for the VMEbus exists in the marketplace. Motorola 68020-based single-board computers will be utilized for 32-bit processing with IEEE floating point coprocessor support. In addition, upward
comparability with 68030-based hardware will be maintained, os-9 has been selected as the operating system, os-9 provides a modular, multiprogramming, multitasking environment with speed for real-time applications and position-independent code. Both C and FORTH will be utilized for programming languages on the LTM with FORTRAN 77, Pascal and Basic also available under os-9 for future users.
3. Summary The LTM design is technically promising for NASA ground-based telerobotic research: The kinematics are unique and highly flexible while being composed of simple mechanisms; hardware for future space applications can be an upgrade of the laboratory units; the arms are modular and allow for easy reconfignration; the capacity-toweight ratio will be excellent compared to existing systems; and the control architecture is expandible and flexible for evolution into autonomous control. Detail design and fabrication of the first LTM Prototype system is underway at ORNL. Detail design is nearing completion, and fabrication will
J.N. Herndon et al. / Robotic Systems for Ground-Based Research
begin shortly. This first prototype will be initially operational by April, 1988.
Acknowledgement The work described in this article represents the combined efforts of a large technical team at ORNL and LaSt. The authors would like to recognize the contributions of S.M. Babcock, H.M. CosteUo, R.L. Glassell, D.P. Kuban, J.C. Rowe, D.M. Williams, and S.D. Zimmermann at ORNL, and R.L. Kurtz and J.E. Pennington at LaRC.
25
Rderences [1] H.L. Martin et al.: "Recommendations for the Next-Generation Space Telerobot System," ORNL/TM-9951 (1986). [2] J.V. Draper, E. Sundstrum, and J.N. Herndon: "Joint Motion Clusters in Servomanipulator Operations," Proceedings of the Human Factors Society 30th Annual Meeting, 1986. [3] S.H. Lowenthal, D.A. Rohn, and B.M. Steinetz: "Application of Traction Drives as Servo Mechanisms," 19th Aerospace Mechanisms Symposium, May 1985. [4] J. Albus: "A Control System Architecture for the Space Station Flight Telerobotic Servicer," Proceedings of the Space Telerobotics Workshop, Jet Propulsion Laboratory, Pasadena, California, January, 1987.
Autonomous Systems for Space
Robotic Systems for NASA Ground-Based Research Joseph N. H e r n d o n and William R. Hamel Oak Ridge National Laboratory *, Oak Ridge, Tennessee 37831, USA
and
Alfred J. Meintel NASA Langley Reseaech Center, Hampton, Virginia 23665, USA The realization of the National Aeronautics and Space Administration (~ASA) Space Station Program will mark the beginning of a major new era in space exploration and utilization. This new era will be identified by much increased utilization of dexterous robotics systems to reduce the need for hazardous extravehicular activities by astronauts. The goals are to improve overall system safety and efficiency as well as provide significant spin-off technology to improve the productivity of the US industrial sector. In support of these goals, the Laboratory Telerobotic Manipulator (LT~) is being developed at ORm. under the direction of IqAS^ Langley Research Center to provide telerobotic hardware for use in NASA ground-based laboratory research. A significant technical limitation is the lack of available telerobotic hardware that can function well as a real-time teleoperator and can also provide a sound hardware basis for intelligent robotic operations. The LTM is being developed to merge these technical domains in common hardware to further NASA'S research investigations. This article summarizes the mechanicul and controls approach being utilized to realize these goals.
1. Introduction
The realization of the National Aeronautics and Space Administration (NASA)Space Station Program will mark the beg0nning of a major new era in space exploration and utilization. This new era will be identified by much increased utilization of dexterous robotics systems to reduce the need for hazardous extravehicular activities by astronauts. The goals are to improve overall system safety and efficiency as well as provide significant spin-off technology to improve the productivity of the us inclustrial sector. In support of these goals, the Laboratory Telerobotic Manipulator (LTM) is being developed at Oak Ridge National Laboratory (ORNL) under the direction of NASALangley Research Center (LaRC)
Joseph N. Herndon is presently Manager of the Laboratory Telerobotic Manipulator Program at Oak Ridge National Laboratory. Mr. Hemdon has over 10 years experience in the fields of remote handling system design and operation, remote facility design, and manipulator research and development. Mr. Herndon has BS and MS degrees in Mechanical
Engineering.
Keywords: Telerobot, Force reflection, Modular, NASA, Traction drives.
William R. ~ is presently Section Head of the Telerobotic Systems Section at Oak Ridge National Laboratory and is a principal investigator in the op.m. Center for Engineering Systems Advanced Research. Dr. Hamel has over 15 years experience in controis and electromechanical systems development. Dr. Hamel has BS, MS and PhD degrees in Mechanical Engineering.
* Research performed at Oak Ridge National Laboratory, operated by Martin Marietta Energy Systems, Inc., for the us. Department of Energy under Contract No. DE-AC05840R21400, and sponsored by the National Aeronautics and Space Administration. North-Holland Robotics 4 (1988) 19-25
Alfred J. Meintei is presently Head of the Automation Technology Branch within the Information Systenas Division at ~ A S A Langley Research Center.
0167-8493/88/$3.50 © 1988, Elsevier Science Publishers B.V. (North-Holland)
20
J.N. Herndon et a L / Robotic Systems for Ground-Based Research
to provide telerobotic hardware for use in NASA ground-based laboratory research. Present NASA plans indicate the need to rely on teleoperation for control of these dexterous systems in the construction and early operation of the Space Station. Evolution to intelligent autonomous robotic operations will be gradual. Due to the unique nature of orbital operations, this evolution must be made in a very carefully controlled and smooth manner. A significant limiting factor in this present technical approach is the lack of available telerobotic hardware that can function well as a real-time teleoperator and can also provide a sound hardware basis for intelligent autonomous robotic operations. The LTM is being developed to form a basis for the merger of these rather diverse technical domains into common hardware to further NASA'S research progress. The LTlvlwill be a highperformance, force-reflecting servomanipulator system to maximize the approach to astronaut EVA task performance while at the same time providing the positioning accuracy capabilities and control architecture necessary for the implementation of sensor-based robotic operations as the evolutionary path toward autonomy. An additional goal is to provide a system concept that minimizes the differences between ground-based research hardware and space systems to aid in demonstrations and development. This overall technical approach is based on the conceptual design study summarized in [1].
2. Technical Approach The merger of the mechanical and control features necessary for a force-reflecting servomanipulator and a robotic positioner into a single system is a particularly difficult task. As shown in Fig. 1, a good force-reflecting servomanipulator designed for efficient human-in-the-loop control emphasizes end effector speed, high joint backdrivability for force reflection, low friction, and low reflected inertia. On the other hand, a good robotic positioner emphasizes end effector accuracy, end effector speeds, and mechanical and control stiffness. As shown in Fig. 1, the LTM design will bridge the gap between these two technologies by providing the most important design and operational parameters of each. The LTM Prototype system will comprise two force-reflecting slave
and master arm pairs with a digital-based control system providing bilateral, position-position, force-reflecting control. In addition, joint-level position and velocity robotic control and openloop joint drives will be provided for implementation of robotic control.
2.1 Mechanical Design The LTM design, as shown in Fig. 2, utilizes a modular approach for joint construction with common pitch-yaw differential joints implemented for the arm shoulder, elbow, and wrist. An output roll follows the wrist pitch-yaw differential to give a three degrees-of-freedom wrist capable of orientation within a compact hemispherical volume [2]. This construction furnishes three degrees of freedom for translational positioning capabilities and an additional redundant degree of freedom which probably will be utilized primarily for translational motion. The benefits of a redundant kinematics approach include multiple solutions for singularity avoidance and improved ability to work in physically constrained areas. This is a proper direction for NASA ground-based research. This approach does result in more possibilities for colinear axes within the arm range-of-motion and will require more complex control techniques both in the teleoperator and in the robotic mode. The LTM WriSt pitch-yaw utilizes a dual differential and traction drives for joint force transmission. The joint design is based on two input drive pinions with one output rotating about two orthogonal axes. The input pinions are driven by DC motors through the differential with variable preloading mechanism to maximize the differential efficiency. A variable preloading mechanism is also incorporated at the joint output. Each motor is equipped with a zero-backlash reducer, brake, tachometer, and optical encoder. Absolute encoders are utiliTed at the pitch-yaw joint outputs to provide high accuracy positioning for robotic operation. A torque sensor is incorporated between the reducer and the input preloading mechanisms. This torque signal will be used to provide improved joint backdrivability for bilateral, force-reflecting teleoperation. The LTM slave shoulder and elbow pitch-yaw joint modules will be similar to the wrist, but will be scaled to a higher capacity to allow for improved perfor-
J.N. Herndon et al. / Robotic Systems for Ground-Based Research
21
GOOD INDUSTRIAL ROBOT
GOOD FORCE-REFLECTING TELEOPERATOR
End effector speed 36 in./s
End effeetor speed 30 to 50 in./s
Friction 1-5% of capacity (at expense of increased backlash)
Friction 30-100% of capacity
Medium to low backlash
No backlash (at expense of increased friction)
Replica master control
Teach pendant, keyboard
I- to 2-in deflection at full load
Minimal deflection at full load (.010 to .05 in.)
6 DOF and end effector
4 to 6 DOF and end effector
Bilateral position-position control for force reflection
Force feedback with 6-axis end effector sensor
Relatively low inertia for minimum fatigue
High inertia for stiffness
Kinematics approximately manlike
Kinematics mission dependent
Accuracy and repeatability not important
Accuracy and repeatability very important
1:4 to i:i0 capacity/weight ratio
i:i0 to 1:40 capacity/weight ratio
Universal end effector
Interchangeable end effeetors
LTM End effector s p e e d ~ 36 in./s Friction close to teleeperator, much lower than robot Backlash close to robot, much lower than teleoperator Replica master control preferable, joysticks and autonomy research possible 0.030-in. deflection at full load Bilateral position-position control for force reflection Low inertia compared to robots Manlike kinematics for dexterity in teleoperation Universal interface for NASA end-effector research Capacity of 20 ib continuous, 30 ib peak Arm cross section to reach inside 6-in.-square opening
Fig. 1. L ~ design basis compared to teleoperator and robot performance.
mance in ground-based operations. The master arms of the LTM will be constructed from three wrist pitch-yaw modules since the required joint capacities can be much lower. Traction drives with preloading mechanisms were chosen for torque transmission through the LTM differentials [3]. Although traction drives have
not been widely used for servocontrol applications to date, they potentially can provide many benefits for these applications, particularly in space. One benefit is no backlash with only local creep occurring at the point of contact. Additional benefits include high torsional stiffness, good efficiencies, and reduced lubrication requirements
22
J.N. Herndon et aL / Robotic Systems for Ground-Based
~/SHOULDERPITCH 5"
Fig. 2. LTM~
~/.ELBONPITCH
layout and modularity.
compared to equivalent capacity gear drives. The penalties for traction drives include decreased volumetric efficiency and use of a technology that is less demonstrated compared to more common techniques such as gears. A full pitch-yaw test
Fig. 3. L ~ pitch-yaw joint test stand.
Research
stand unit has been fabricated at ORNL to begin development of control algorithms and to verify the traction drive design methodology. This first test stand is illustrated in Fig. 3 with the traction drive differential shown in a close-up view in Fig. 4. This test stand is presently operational at ORNL and a second test stand is nearing completion. The second test stand will allow master-slave control algorithm development in a controlled bench-top environment prior to implementation on the full LTM system. Finally, the mechanical joint concept was developed to provide a cost-effective solution for NASA laboratory research needs. A significant number of follow-on LTM systems are presently planned. The common joint design concept lends itself to reduced design and fabrication costs. Commercially available componentry is being utilized to the maximum extent practical. The entire design and fabrication effort at ORNL will utilize computer-integrated manufacturing techniques to minimize fabrication costs and to ease the NASA technology transfer process.
23
J.N. Herndon et al. / Robotic Systems for Ground-BasedResearch 2.2 Control System
Fig. 4. LrMjoint differentialand tractiondrives.
Development of the LTM control system will be a significant challenge. ORNL brings sitmificant experience in the development and implementation of digital control systems for force-reflecting servomanipulators gained through successful hardware and software development on three separate servomanipulator systems over the past six years. In addition to the control modes described earlier, the LTM control system must provide interfaces for dextrous end effectors and must provide "hooks" for alternate slave arm controllers such as universal hand controllers, joystick resolved rate controllers, optimized replica controllers, and external robotic/autonomous controllers. It is of utmost importance that the LTM control system architecture be flexible and allow layering into hierarchical control architectures such as the NASA/CNationalBureau of Standards NASREM architecture [4]. Fig. 5 illustrates the functional approach to implementing a digital-based control system for bilateral, force-reflecting control while prodding the embedded needs for implementation of intelligent robotic control techniques. The shaded blocks
Path Plenner end Sensory Transformer
Force/Torque
World
Sensory Feedback System
Fig. 5. Servomanipulator/intelligentrobot functionalcontroldiagram.
24
J.N. Herndon et al. / Robotic Systems for Ground-Based Research
FUTURELINKTO UPPERLEVELSIN HIERARCHICALSYSTEM
I..
Interface~ - ~
3
I
Fiber Optic CommunicationLink
LEFT ARM SHOULDER
RIGHTARM
~
J SHOULDER
LEFT ARM
RIGHTARM
SHOULDER L
i SHOULDER i
ELBOW WRIST
~
-! ELBOW
ELBOW
J
WRIST
~ WRIST
WRIST ROLL/ END EFF. |
vI ENDEFF.
WRIST
WRIST ROLL/ END EFF.
I
ELBOW
I
WRIST ROLL/
~Distributed OnboardElectronics( Typical )
Fig.6.LTMcontrolsystemhardwareblockdiagram. in Fig. 5 are those actually being implemented initially on the LTM Prototype system. The LTM control system hardware block diagram is shown in Fig, 6. As described earlier, the LTM will have significant numbers of signals originating within each pitch-yaw joint module. In order to reduce the arm onboard cabling bundle to a manageable size, all onboard sensor data will be locally collected within each joint module and processed onto a serial link for communication to the central control rack. The onboard data collection and communication hardware will be of custom design for this application. As shown in Fig. 6, each of the master and slave arm pairs will be provided with a central control rack. Master-to-slave and slave-to-master communication will also be handled by high-speed serial communication (10 megabits per second) over a fiber optic link. The central control racks for the master and slave will each be based on the vMEbus backplane for an open architecture on an industry standard (IEEE P1014) bus structure with rugged construction. Also, much vendor and product support for the VMEbus exists in the marketplace. Motorola 68020-based single-board computers will be utilized for 32-bit processing with IEEE floating point coprocessor support. In addition, upward
comparability with 68030-based hardware will be maintained, os-9 has been selected as the operating system, os-9 provides a modular, multiprogramming, multitasking environment with speed for real-time applications and position-independent code. Both C and FORTH will be utilized for programming languages on the LTM with FORTRAN 77, Pascal and Basic also available under os-9 for future users.
3. Summary The LTM design is technically promising for NASA ground-based telerobotic research: The kinematics are unique and highly flexible while being composed of simple mechanisms; hardware for future space applications can be an upgrade of the laboratory units; the arms are modular and allow for easy reconfignration; the capacity-toweight ratio will be excellent compared to existing systems; and the control architecture is expandible and flexible for evolution into autonomous control. Detail design and fabrication of the first LTM Prototype system is underway at ORNL. Detail design is nearing completion, and fabrication will
J.N. Herndon et al. / Robotic Systems for Ground-Based Research
begin shortly. This first prototype will be initially operational by April, 1988.
Acknowledgement The work described in this article represents the combined efforts of a large technical team at ORNL and LaSt. The authors would like to recognize the contributions of S.M. Babcock, H.M. CosteUo, R.L. Glassell, D.P. Kuban, J.C. Rowe, D.M. Williams, and S.D. Zimmermann at ORNL, and R.L. Kurtz and J.E. Pennington at LaRC.
25
Rderences [1] H.L. Martin et al.: "Recommendations for the Next-Generation Space Telerobot System," ORNL/TM-9951 (1986). [2] J.V. Draper, E. Sundstrum, and J.N. Herndon: "Joint Motion Clusters in Servomanipulator Operations," Proceedings of the Human Factors Society 30th Annual Meeting, 1986. [3] S.H. Lowenthal, D.A. Rohn, and B.M. Steinetz: "Application of Traction Drives as Servo Mechanisms," 19th Aerospace Mechanisms Symposium, May 1985. [4] J. Albus: "A Control System Architecture for the Space Station Flight Telerobotic Servicer," Proceedings of the Space Telerobotics Workshop, Jet Propulsion Laboratory, Pasadena, California, January, 1987.