- Email: [email protected]
Homonoia: When your Car Reads your Mind
Available online at www.sciencedirect.com Available online at www.sciencedirect.com
Available online at www.sciencedirect.com
ScienceDirect Procedia Computer Science (2016) 000–000 Procedia Computer Science 11000 (2017) 135–142 Procedia Computer Science 00 (2016) 000–000
www.elsevier.com/locate/procedia www.elsevier.com/locate/procedia
The The 14th 14th International International Conference Conference on on Mobile Mobile Systems Systems and and Pervasive Pervasive Computing Computing (MobiSPC 2017) (MobiSPC 2017)
Homonoia: Homonoia: When When your your Car Car Reads Reads your your Mind Mind a a,∗ a Ahmad Ahmad Bennakhi Bennakhia ,, Maytham Maytham Safar Safara,∗,, Jafar Jafar Abdulrasoul Abdulrasoula a Kuwait a Kuwait
University, Kuwait City, Safat PO Box 5969, Kuwait University, Kuwait City, Safat PO Box 5969, Kuwait
Abstract Abstract This paper aims to shed light on the possibility of applying brain waves as a mean to understand the driver’s mood while on This paper aims to shed light on the possibility of applying brain waves as a mean to understand the driver’s mood while on his/her everyday commutes. While there are several studies that document the relationship between brain waves and mood, none his/her everyday commutes. While there are several studies that document the relationship between brain waves and mood, none have progressed to apply it on vehicles. We experimented the use of EEG sensors to detect brain waves and recorded several have progressed to apply it on vehicles. We experimented the use of EEG sensors to detect brain waves and recorded several correlations that could prove to be useful. After understanding the driver’s mood only through a passive approach, the car could correlations that could prove to be useful. After understanding the driver’s mood only through a passive approach, the car could suggest ways in which it could improve or compliment the driver’s mood. This could also open up a whole level of danger diversion suggest ways in which it could improve or compliment the driver’s mood. This could also open up a whole level of danger diversion features in cars, if both the car and the EEG sensor are integrated well enough. features in cars, if both the car and the EEG sensor are integrated well enough. © 2016 The Authors. Published by Elsevier B.V. © 2017 2016 The TheAuthors. Authors.Published Publishedby byElsevier ElsevierB.V. B.V. © Peer-review under responsibility of the Conference Program Chairs. Peer-review under responsibility of the Conference Program Chairs. Keywords: Keywords: BCI, Car Interface, Mood sensing, Ambient Technology, Contextual Navigation, EEG, Smart Systems BCI, Car Interface, Mood sensing, Ambient Technology, Contextual Navigation, EEG, Smart Systems
1. Introduction 1. Introduction Homonoia is defined as “oneness of mind, unanimity, and concord” 11 . This word describes the relationship that Homonoia is defined as “oneness of mind, unanimity, and concord” . This word describes the relationship that we human beings are trying to achieve with our computers. The gap between computers and human beings is still a we human beings are trying to achieve with our computers. The gap between computers and human beings is still a wide one, but is closing up pretty quickly considering the annual semi-exponential growth of transistor density and the wide one, but is closing up pretty quickly considering the annual semi-exponential growth of transistor density and the ubiquitous presence of advanced microcontrollers. Sight (screens and cameras), sound, and touch are all integrated in ubiquitous presence of advanced microcontrollers. Sight (screens and cameras), sound, and touch are all integrated in our daily life to the point that we almost take it for granted with every smart phone released. However, brain waves our daily life to the point that we almost take it for granted with every smart phone released. However, brain waves are not truly well integrated in the practical course of our lives considering that they potentially could tell us more are not truly well integrated in the practical course of our lives considering that they potentially could tell us more about our well-being more than any other parameter, especially while driving. about our well-being more than any other parameter, especially while driving. Using brain waves in a car is not entirely a new concept since it has been used to control cars 22 33 . These applications Using brain waves in a car is not entirely a new concept since it has been used to control cars . These applications are indeed very helpful for handicapped people and could be used in the future for all people. Nevertheless, it is a are indeed very helpful for handicapped people and could be used in the future for all people. Nevertheless, it is a long way from industrial scale implementation and will likely die out since driverless cars are the wave that is going long way from industrial scale implementation and will likely die out since driverless cars are the wave that is going to topple it all. Our study is not about driving cars using your brain waves, rather it is about making the computer to topple it all. Our study is not about driving cars using your brain waves, rather it is about making the computer interface in your car a travel companion that can sense your concentration, mood, and dangerous conditions such as interface in your car a travel companion that can sense your concentration, mood, and dangerous conditions such as ∗ ∗
Maytham Safar. Tel.: +965-99361062 Fax: +965-24839461 Maytham Safar. Tel.: +965-99361062 Fax: +965-24839461 E-mail address: [email protected] E-mail address: [email protected]
1877-0509 © 2016 The Authors. Published by Elsevier B.V. 1877-0509 © 2016 The Authors. Published by Elsevier B.V. Peer-review under responsibility of the Conference Program Chairs. Peer-review©under of the Conference Program B.V. Chairs. 1877-0509 2017responsibility The Authors. Published by Elsevier Peer-review under responsibility of the Conference Program Chairs. 10.1016/j.procs.2017.06.134
136 2
Ahmad Bennakhi et al. / Procedia Computer Science 110 (2017) 135–142 A. Bennakhi et al. / Procedia Computer Science 00 (2016) 000–000
drowsiness, drunkenness, or an eminent epileptic seizure. Implementing this would require no safety procedure since the worst case scenario in all of that is just a misplaced suggestion. There exists an enormous amount of studies that map every state of mind to its corresponding alterations in brain wave signals, but never has there been one to question the application of introducing brain waves to ambient intelligence presence inside cars instead of only controlling the car itself. Sensing brainwaves can also have a tremendous boost to the accuracy contextual navigation and, if perfected, can be an AI present in cars that will make cars a place where interface and user can interact tunefully. Section 1 will be the introduction, while section 2 will discuss studies related to this paper. Section 3 will give the reader some background knowledge about both the topics of brainwaves and the state of mind corresponding to it. Moving on to section 4, we will discuss the details of the case study that this paper is going to tackle. We also will display and analyze the results of the case study with all of its implications in section 4 too. A design of the proposed system that integrates the EEG device with the car is proposed in section 5 with all of its potential applications. The study finally concludes in section 6 with a concise summary of our important findings. Nomenclature Delta waves Theta waves Alpha waves Beta waves Gamma waves
Between 1 Hz and 4 Hz Between 4 Hz and 8 Hz Between 7.5 Hz and 13 Hz Between 13 Hz and 30 Hz Between 30 Hz and 44 Hz
2. Related Work A previous study 4 went through the idea of predicting distractions, using a system that the researchers have programmed to detect. What we plan to do differently from the previously mentioned study is to establish the foundations of a mood sensing system that would not only sense the mood, but also the various mental conditions that could affect the driver using real life driving data (the mentioned study used a computer driving simulator to develop and test their system). Truck drivers were subject of continuous research when it came to driving and EEG. Most of them focused on fatigue/sleepiness during long distance driving 5 . While this paper is old (1993), it marked the beginning of research that focuses on the brain waves of drivers during their journey. More papers emerged with astounding results when it came to predicting the driver’s drowsiness/fatigue 6 7 . All of the previously mentioned studies were done using simulated driving rather than real life testing. A survey was done archiving and comparing all of the studies concerned the detection of drowniness and fatigue while driving, whether it was done using EEG or by other means 8 . It is certainly a good read if someone wants to be informed of the research that was done on the field before 2009. A study 9 had also been conducted to link brain cognition to the mood of individuals. The study also uses an EEG sensor to verify their hypothesis and record the correlation. Predicting the mood while walking was predicated in a study 10 that used electromyogram(EMG) sensors all over the body. Our method is not as intrusive, as it only requires the user to wear a comfortable headband. The use of EEGs to predict alertness and mood was present in a study 11 that wanted to detect the effect of colors on brain waves and concentration. The experiment was not performed on a driver though, since its scope was not specified for that. 3. EEG Brain waves Brain waves have been a under research since the beginning of the 20th century, and they have been thoroughly studied until our present day. A lot of the work has been done on brain waves that researchers can pinpoint which wave lengths correspond with each state of mind. Moreover, the effects of numerous psychiatric medications have been known to produce anticipated brain wave abnormalities. Today’s market offers us a wide range of EEG sensors
Ahmad Bennakhi et al. / Procedia Computer Science 110 (2017) 135–142 A. Bennakhi et al. / Procedia Computer Science 00 (2016) 000–000
137 3
that impose as little discomfort as wearing glasses. As this technology becomes more ubiquitous we will examine the potential applications of the technology while driving a vehicle. The brain waves classification is a nomenclature, especially when it comes to the boundary of each band. Our EEG headband can detect 5 bands, therefore we based our classification of the EEG spectrum on the headband that was used in our study and listed in the nomenclature. The following brain waves are listed according to what the majority of sources consider them to be. The delta band is the lowest one of these wave bands (below 4 Hz). It is mainly associated with deep dreamless sleep, but is also present when a person is doing tasks that require continuous attention for a long period of time 12 . Theta waves (4 Hz to 8 Hz) on the other hand is present in REM sleep, meditation, drowsiness, idling, and repression of a certain action or response 12 . The Alpha (8 Hz to 13 Hz) band is related to relaxation, daydreaming, and closed eyes. In an activity such as watching TV, alpha waves are often the dominant band. Beta waves (13 Hz to 30 Hz) signify alertness, focus, and anxiousness. Beta waves could range from a calm active state of mind to intense or stress to mild obsessiveness. The last band on the list is Gamma (30+ Hz), which is associated with hyper alertness and high learning. Gamma waves are evident during cross-modal sensory processing and when a person is thinking very deeply looking for insights 13 . The following points can be summarized in table 1 (shown below). Table 1. EEG bands and the state of mind that are associated with
Band
Frequency (t)
Associated state of mind (t)
Delta Theta Alpha Beta Gamma
Less than 4 Hz 4 Hz to 8 Hz 8 Hz to 15 Hz 16 Hz to 31 Hz More than 31 Hz
Deep dreamless sleep or K-complex 14 15 REM, meditation 16 , drowsiness 17 , idling or repression of action/response 18 Daydreaming, relaxed, being awake with eyes closed 15 Awake, alert, engaged, focused 19 , or anxious 20 Hyper alert and cross-modal sensory processing (perception that combines two different senses, such as sound and sight) 21
Some books have even described personality types that have tendencies to dominate certain bands 22 . With all of that information flowing, we are going to talk about the technicality of harvesting it in a particular case study and applying it in a useful way. 4. Case Study: Mood Detection while Driving 4.1. Method The brain wave headset that we used in this case study is the Muse™ headband. Muse™ has 7 sensors that are strategically placed on the forehead, temple, and behind the ears. We kept in mind that the headset has to be comfortable and reliable at the same time, that is why we chose Muse™ in particular. The headband also has an adequate library that makes coding a custom application much easier. The experiment begins with opening a custom mobile application that was created specifically for this study. This mobile application syncs with the headset placed on the volunteer’s head. After a connection is made between the headband and the mobile application, the session begins with a simple question of “How are you feeling?” with 7 options to choose from. After doing so, the volunteer drives along his intended route in a regular fashion without being distracted by anything related to the experiment. Once the volunteer reaches his destination, he would end the session and answer three more questions. The first being “How are you feeling?” with the reply options that were available earlier on, and the second question is to determine the level of traffic by selecting from 3 choices. The last question will ask if the user telephoned anyone during his trip. All of the test subjects are over 18 and live in Kuwait City, which is considered a big metropolitan city. All sessions recorded are over 5 minutes long and have adequate signal strength from the headset. The application that is being used is installed in smartphones that are given to the volunteers. A total of 11 volunteers(6 males and 5 females)
Ahmad Bennakhi et al. / Procedia Computer Science 110 (2017) 135–142
138 4
A. Bennakhi et al. / Procedia Computer Science 00 (2016) 000–000
Figure 1. A participant using the EEG device
participated in this experiment and each were asked to perform a total of 5 sessions. It is worth mentioning that some sessions were discarded due to the poor quality and device misuse, cumulatively the total number of sessions is 44. The sessions were all recorded by the Muse™ headband, and the parameter that was used in our study was the absolute band power. The absolute band power for a given frequency range is the logarithm of the sum of the Power Spectral Density of the EEG data over that frequency range. They are provided for each of the four to six channels/electrode sites on Muse 23 . The maximum power recorded out of all of the six electrodes was taken as the most valid value, since it is the strongest signal detected from the brain at the specific frequency band. The data coming from the headband would be flagged in the following way: • If the mood in the beginning matches the mood in the end of his/her ride then the whole session would be flagged in the mood specified by the driver as it is assumed that there were no major mood changes through their journey. • If the mood recorded in the beginning does not match the mood in the end, then the first 1000 (the average interval between transfers is 0.1 seconds) transfers of brain wave data are flagged with the mood recorded in the beginning of the session. Similarly, the last 1000 transfers are flagged with the mood recorded at the end of the session.
4.2. Results The sessions were analyzed using the bivariate correlation method: Kendall’s tau-b. The data was first separated into different files according to mood then analyzed separately for correlations. The variables that were analyzed were the brain wave bands: Alpha, Beta, Delta, Theta, and Gamma. The result was 7 correlation tables(each mood had a table). The combined results of these tables can be seen in table 2. The correlations that were significantly positive are listed beside a plus sign, while the negative ones are listed beside a negative dash sign. The results that yielded correlations that were close to zero are listed beside bullet points. The Alpha-Beta correlation coefficient is high (more than doubled) for non-negative emotions such as neutrality, happiness, and tired. While sadness shows a bit of a significant inverse correlation. Moreover, sadness varies from the other emotions in the case that it is the only emotion with a relatively high Alpha-Delta correlation. The AlphaTheta band also has relatively high correlation coefficients when it comes to emotions such as hunger and sadness. The Alpha-Gamma correlation is unique since it has a relatively high inverse correlation coefficient when it comes to happiness and sadness, which are an unlikely pair even when compared with the other band correlations in this study. The Alpha-Gamma correlation is highest on the positive side when people reported feeling “neutral”, which may suggest that emotions whether happy or sad contribute to a negative correlation coefficient, and lack of them contribute to the positive end. Negative emotions such as anger, stress, hunger, and sadness all have relatively high inverse Beta-Delta correlation coefficients especially anger. Beta also has a statistically significant inverse correlation with Theta when people recorded being stressed and angry. Across all emotions, the Beta-Gamma correlation coefficients are highly positive with somber moods such as being tired and sad, ranking the lowest. Higher positive correlation coefficients are also present between the Delta-Theta bands with stress and being tired, having the highest of them all. The list of negative emotions (hunger, anger, and sadness) show common features once again by having similar negative correlation
Ahmad Bennakhi et al. / Procedia Computer Science 110 (2017) 135–142 A. Bennakhi et al. / Procedia Computer Science 00 (2016) 000–000
139 5
Table 2. The correlation coefficient is written inside the brackets of each mood. Highlighted results are deemed to be of interest.
Alpha
Beta +Neutral(0.308), Tired (0.308), Happy ( 0.3) Hungry( 0.142), Stressed(0.109) •Angry( -0.068) - Sad( -0.132)
Beta
Delta
Theta
Delta +Sad(0.248) •Neutral(0.082), Tired(0.072), Hungry(0.046), Angry(0.019), Happy(0.013) , Stressed(-0.054) +Happy(0.077) •Tired(0.018), Neutral(-0.023), -Stressed(-0.111), Hungry(-0.113), Sad(-0.181), Angry(-0.242)
Theta +Sad(0.232), Hungry(0.198), Tired(0.144) • Happy(0.088), Neutral(0.076), Angry(0.06), Stressed(-0.006) +Sad(0.14), Tired(0.109) • Hungry(0.018), Happy(-0.008), Neutral(-0.022) -Stressed(-0.165), Angry(-0.197) +Stressed(0.395), Tired(0.319), Hungry(0.266), Happy(0.240), Angry(0.234), Neutral(0.184), •Sad(0.058)
Gamma +Neutral(0.165), Tired(0.096), Hungry(0.084), Stressed(0.064) •Angry(-0.002) -Sad(-0.268), Happy(-0.28) +Hungry(0.605), Stressed(0.589), Neutral(0.535), Happy(0.517), Angry(0.501), Tired(0.495), Sad(0.317) • Happy(0.018), Tired(0.014), Neutral(-0.04), Stressed(-0.066) -Hungry(-0.165), Angry(-0.192), Sad(-0.242) +Tired(0.111) •Sad(0.081), Neutral(-0.019), Happy(-0.022), Hungry(-0.061) -Stressed(-0.172), Angry(-0.312)
coefficients between the Delta-Gamma bands. Finally, the Theta-Gamma correlation coefficients show an intense anger mood on the significant negative end and a worn-out tired mood on the positive end of the correlation table. In between several brain wave bands, we can notice a set of emotions that mostly go together. Most noticeably sadness and hunger tend to have similar correlations. Less negative moods such as being happy, tired, and neutral have also noted to stick together. That being said, there are several cross band correlations that can act as indicators to point out the mood that the driver is feeling. For example, analyzing the Alpha-Gamma band can reveal if the person is neutral or happy, both of which normally stick together but differ greatly in this case. 4.3. Implications This is just a preliminary proof of concept study and it already produced results that could be used in real life applications. Only one type of correlation method was applied to the data, so further analysis would no doubt reveal more interesting data. A car interface application that monitors and compares bandwidths would definitely be able to differentiate between moods given that the subject is researched more thoroughly with more participants. Moreover, if the car application is programmed to adjust to the driver’s personal brain wave patterns, then an even more concise mood recognition system could be developed. Mood recognition using EEG is a fairly new area that is being explored. So if this system can also be implemented in autonomous cars then an array of new possibilities will open up for new applications that can utilize an EEG headband. The user of such a system could be monitored not just for mood, but other conditions as well such as
Ahmad Bennakhi et al. / Procedia Computer Science 110 (2017) 135–142 A. Bennakhi et al. / Procedia Computer Science 00 (2016) 000–000
140 6
epileptic seizures, fatigue or alcohol abuse 24 . These areas are drenched with already existing research studies that would help build a basis for a system that would detect these conditions. 5. Proposed System Design Before we begin to outline how the system works, here are the guidelines that will be the basis of our proposed system: 1. The system will not commit to any action unless the user gives his/her consent (if the driver has the capability of making decisions). 2. The system should not have any control over the car’s driving system unless autonomous driving becomes approved by the concerned government associations. 3. The system should impose minimal or no distraction at all on the driver. 4. The driver should not have to order the system to do anything (the system has to anticipate his needs), since it is purely based on the system giving suggestions or asking for consent to perform a task. A brain wave sensor has to be used so that the system can be implemented. The brain wave headband can have auxiliary parts that make it have more than one particular function (e.g. glasses, Bluetooth headset, or a visor). As seen in the figure below, the sensor that we used in our case study can be an integral part of many accessories. Therefore, it would be easy for users to wear it casually and still reap the benefits of the implemented system. There are two flavors of Homonoia: Table 3. The flavors of Homonoia
Flavor
Can Access
Requirements
Skinny
Car media player, Smartphone, Car Interface, and GPS
An integrated application that is installed in the car’s interface system
Autonomous
In addition to the skinny flavor control privileges, it also has control over the car seat and driving system
In addition to the skinny flavor requirements, it also requires a fully secure autonomous driving system that would authorize emergency intervention when needed.
5.1. Skinny Self-driving cars are in no way fail-proof. A report 25 compiled by the DMV summarized the severity of the situation by documenting the total of times that self-driving cars disengaged, whether the cause of the disengagement was a hardware/software malfunction or just the driver sensing that something went wrong. That is why we need an implementation that would isolate the car controls from the Homonoia system. The skinny flavor would only have control over the car’s interface and smartphone only. It should also be clear that smartphone access is purely optional, but will also make the integration of the system much more comprehensive and efficient. The skinny Homonoia flavor has a unilateral sequence of action: Recognition: While the headset is forwarding EEG data to the car’s interface system, there is a constant search for parameters to recognize the mood (e.g. angry, happy...etc.) . Interaction: A simple relevant question is directed towards the user so that an action can be commenced (e.g. “Do you want me to turn on the soothing music?” Or “Theres a caf nearby, do you want to grab a cup of coffee from there?”). The user would respond by a yes, no, or another reaction that would be recognized by the natural language voice recognition system.
Ahmad Bennakhi et al. / Procedia Computer Science 110 (2017) 135–142 A. Bennakhi et al. / Procedia Computer Science 00 (2016) 000–000
141 7
Figure 2. The proposed system summary
Recording: All replies from the user is documented as well as the time and other variables that could affect the circumstances, so that future suggestions could become adaptively better. The system is bound to get better with each suggestion, so that the number of hits per suggestion would steadily increase, eventually reaching homonoia status. 5.2. Autonomous In addition to all of the functions that the skinny flavor offers, the Autonomous flavor offers an emergency response option. There are numerous studies that demonstrated the ability to detect mental conditions such as encephalopathy, strokes, sleep disorders, disorientation, alcohol abuse, epilepsy, and many other brain abnormalities 24 . The profile for each brain condition can be saved and referenced so that Homonoia checks for its occurrence. This flavor can be implemented when self-driving cars reach a sufficient level of safety. In case a brain wave abnormality is detected, the autonomous driving system would be placed on alert mode. Being on alert mode still keeps the driver in control of the car, but stays on the look for driving hazards such as collisions or swaying out of the road. Furthermore, a timed enquiry will be posted on the cars interface to make sure if the driver is in good mental condition. If a user manages to respond with an appropriate answer, only then that car would return to normal mode. Otherwise, if the driver did not respond, or a dangerous driving pattern was detected the car would transition to panic mode. The vehicle would have to either park on the side of the road or drive autonomously to the nearest hospital. The driver’s seat would also be adjusted to match the condition suitable for the user’s brain abnormality. 6. Conclusion We have exhibited how brain waves can act as indicators of numerous brain states. The previous studies that integrated brain waves with technology have focused mainly on it having an active role when driving, which needs a
142 8
Ahmad Bennakhi et al. / Procedia Computer Science 110 (2017) 135–142 A. Bennakhi et al. / Procedia Computer Science 00 (2016) 000–000
lot of precaution and may even be rendered obsolete with the emergence of autonomous cars. Therefore, we decided to examine the passive role that brain waves can take to compliment other technologies rather than compete with them. We have carried out an experiment that could demonstrate whether we could harness certain correlation patterns of brain wave bands to pin point the driver’s mood. The results show that certain negative moods usually group together with similar correlation coefficients between several brain wave bands. We have also proposed a system that acts in a way that adds minimal workload on the user while integrating him/her as much as possible with an intelligent car interface system. Of course, more research needs to be done in this area in order for this system to be applied to reality in an efficient way. Acknowledgements Special thanks go to Ali Al-Hasan for providing us with a number of volunteers willing to take part in the experiment. References 1. West, W.C.. Hellenic homonoia and the new decree from plataea. Greek, Roman and Byzantine Studies 1977;18(4):307. 2. Chinese researchers unveil brain powered car. 2015. URL: http://www.reuters.com/article/ us-china-brainpower-car-idUSKBN0TQ23620151207. 3. Braindriver. 2015. URL: http://autonomos-labs.com/projects/past-projects/braindriver/. 4. Wang, S., Zhang, Y., Wu, C., Darvas, F., Chaovalitwongse, W.A.. Online prediction of driver distraction based on brain activity patterns. IEEE Transactions on Intelligent Transportation Systems 2015;16(1):136–150. 5. Kecklund, G., Åkerstedt, T.. Sleepiness in long distance truck driving: an ambulatory eeg study of night driving. Ergonomics 1993;36(9):1007–1017. 6. Lal, S.K., Craig, A., Boord, P., Kirkup, L., Nguyen, H.. Development of an algorithm for an eeg-based driver fatigue countermeasure. Journal of safety Research 2003;34(3):321–328. 7. Kim, J.Y., Jeong, C.H., Jung, M.J., Park, J.H., Jung, D.H.. Highly reliable driving workload analysis using driver electroencephalogram (eeg) activities during driving. International Journal of Automotive Technology 2013;14(6):965–970. 8. Liu, C.C., Hosking, S.G., Lenn´e, M.G.. Predicting driver drowsiness using vehicle measures: Recent insights and future challenges. Journal of safety research 2009;40(4):239–245. 9. Yuan, J., Xu, S., Yang, J., Liu, Q., Chen, A., Zhu, L., et al. Pleasant mood intensifies brain processing of cognitive control: Erp correlates. Biological psychology 2011;87(1):17–24. 10. Kinase, Y., Venture, G.. Estimating mood variation from mpf of emg during walking. In: Engineering in Medicine and Biology Society (EMBC), 2013 35th Annual International Conference of the IEEE. IEEE; 2013, p. 4981–4984. 11. Plitnick, B., Figueiro, M., Wood, B., Rea, M.. The effects of red and blue light on alertness and mood at night. Lighting Research & Technology 2010;42(4):449–458. 12. Kirmizi-Alsan, E., Bayraktaroglu, Z., Gurvit, H., Keskin, Y.H., Emre, M., Demiralp, T.. Comparative analysis of event-related potentials during go/nogo and cpt: decomposition of electrophysiological markers of response inhibition and sustained attention. Brain research 2006;1104(1):114–128. 13. Kisley, M.A., Cornwell, Z.M.. Gamma and beta neural activity evoked during a sensory gating paradigm: effects of auditory, somatosensory and cross-modal stimulation. Clinical neurophysiology 2006;117(11):2549–2563. 14. De Gennaro, L., Ferrara, M., Bertini, M.. The spontaneous k-complex during stage 2 sleep: is it the forerunnerof delta waves? Neuroscience letters 2000;291(1):41–43. 15. Marcuse, L.V., Fields, M.C., Yoo, J.J.. Rowan’s Primer of EEG. Elsevier Health Sciences; 2015. 16. Kubota, Y., Sato, W., Toichi, M., Murai, T., Okada, T., Hayashi, A., et al. Frontal midline theta rhythm is correlated with cardiac autonomic activities during the performance of an attention demanding meditation procedure. Cognitive Brain Research 2001;11(2):281–287. 17. Makeig, S., Jung, T.P.. Tonic, phasic, and transient eeg correlates of auditory awareness in drowsiness. Cognitive Brain Research 1996;4(1):15–25. 18. Di Lazzaro, V., Pilato, F., Saturno, E., Oliviero, A., Dileone, M., Mazzone, P., et al. Theta-burst repetitive transcranial magnetic stimulation suppresses specific excitatory circuits in the human motor cortex. The Journal of physiology 2005;565(3):945–950. 19. Dustman, R.E., Boswell, R.S., Porter, P.B.. Beta brain waves as an index of alertness. Science 1962;137(3529):533–534. 20. Hammond, D.C.. Neurofeedback with anxiety and affective disorders. Child and adolescent psychiatric clinics of North America 2005;14(1):105–123. 21. Jemmer, P.. Getting in a (brain-wave) state through entrainment, meditation and hypnosis. Hypnotherapy Journal 2009;2:24–29. 22. Kropotov, J.D.. Quantitative EEG, event-related potentials and neurotherapy. Academic Press; 2010. 23. Available data. 2017. URL: http://developer.choosemuse.com/research-tools/available-data#Absolute_Band_Powers. 24. Eeg: Medlineplus medical encyclopedia. 2017. URL: https://www.nlm.nih.gov/medlineplus/ency/article/003931.htm. 25. Vehicles, C.D.o.M.. Autonomous vehicle disengagement reports 2016. 2016. URL: https://www.dmv.ca.gov/portal/dmv/detail/ vr/autonomous/disengagement_report_2016.
Available online at www.sciencedirect.com
ScienceDirect Procedia Computer Science (2016) 000–000 Procedia Computer Science 11000 (2017) 135–142 Procedia Computer Science 00 (2016) 000–000
www.elsevier.com/locate/procedia www.elsevier.com/locate/procedia
The The 14th 14th International International Conference Conference on on Mobile Mobile Systems Systems and and Pervasive Pervasive Computing Computing (MobiSPC 2017) (MobiSPC 2017)
Homonoia: Homonoia: When When your your Car Car Reads Reads your your Mind Mind a a,∗ a Ahmad Ahmad Bennakhi Bennakhia ,, Maytham Maytham Safar Safara,∗,, Jafar Jafar Abdulrasoul Abdulrasoula a Kuwait a Kuwait
University, Kuwait City, Safat PO Box 5969, Kuwait University, Kuwait City, Safat PO Box 5969, Kuwait
Abstract Abstract This paper aims to shed light on the possibility of applying brain waves as a mean to understand the driver’s mood while on This paper aims to shed light on the possibility of applying brain waves as a mean to understand the driver’s mood while on his/her everyday commutes. While there are several studies that document the relationship between brain waves and mood, none his/her everyday commutes. While there are several studies that document the relationship between brain waves and mood, none have progressed to apply it on vehicles. We experimented the use of EEG sensors to detect brain waves and recorded several have progressed to apply it on vehicles. We experimented the use of EEG sensors to detect brain waves and recorded several correlations that could prove to be useful. After understanding the driver’s mood only through a passive approach, the car could correlations that could prove to be useful. After understanding the driver’s mood only through a passive approach, the car could suggest ways in which it could improve or compliment the driver’s mood. This could also open up a whole level of danger diversion suggest ways in which it could improve or compliment the driver’s mood. This could also open up a whole level of danger diversion features in cars, if both the car and the EEG sensor are integrated well enough. features in cars, if both the car and the EEG sensor are integrated well enough. © 2016 The Authors. Published by Elsevier B.V. © 2017 2016 The TheAuthors. Authors.Published Publishedby byElsevier ElsevierB.V. B.V. © Peer-review under responsibility of the Conference Program Chairs. Peer-review under responsibility of the Conference Program Chairs. Keywords: Keywords: BCI, Car Interface, Mood sensing, Ambient Technology, Contextual Navigation, EEG, Smart Systems BCI, Car Interface, Mood sensing, Ambient Technology, Contextual Navigation, EEG, Smart Systems
1. Introduction 1. Introduction Homonoia is defined as “oneness of mind, unanimity, and concord” 11 . This word describes the relationship that Homonoia is defined as “oneness of mind, unanimity, and concord” . This word describes the relationship that we human beings are trying to achieve with our computers. The gap between computers and human beings is still a we human beings are trying to achieve with our computers. The gap between computers and human beings is still a wide one, but is closing up pretty quickly considering the annual semi-exponential growth of transistor density and the wide one, but is closing up pretty quickly considering the annual semi-exponential growth of transistor density and the ubiquitous presence of advanced microcontrollers. Sight (screens and cameras), sound, and touch are all integrated in ubiquitous presence of advanced microcontrollers. Sight (screens and cameras), sound, and touch are all integrated in our daily life to the point that we almost take it for granted with every smart phone released. However, brain waves our daily life to the point that we almost take it for granted with every smart phone released. However, brain waves are not truly well integrated in the practical course of our lives considering that they potentially could tell us more are not truly well integrated in the practical course of our lives considering that they potentially could tell us more about our well-being more than any other parameter, especially while driving. about our well-being more than any other parameter, especially while driving. Using brain waves in a car is not entirely a new concept since it has been used to control cars 22 33 . These applications Using brain waves in a car is not entirely a new concept since it has been used to control cars . These applications are indeed very helpful for handicapped people and could be used in the future for all people. Nevertheless, it is a are indeed very helpful for handicapped people and could be used in the future for all people. Nevertheless, it is a long way from industrial scale implementation and will likely die out since driverless cars are the wave that is going long way from industrial scale implementation and will likely die out since driverless cars are the wave that is going to topple it all. Our study is not about driving cars using your brain waves, rather it is about making the computer to topple it all. Our study is not about driving cars using your brain waves, rather it is about making the computer interface in your car a travel companion that can sense your concentration, mood, and dangerous conditions such as interface in your car a travel companion that can sense your concentration, mood, and dangerous conditions such as ∗ ∗
Maytham Safar. Tel.: +965-99361062 Fax: +965-24839461 Maytham Safar. Tel.: +965-99361062 Fax: +965-24839461 E-mail address: [email protected] E-mail address: [email protected]
1877-0509 © 2016 The Authors. Published by Elsevier B.V. 1877-0509 © 2016 The Authors. Published by Elsevier B.V. Peer-review under responsibility of the Conference Program Chairs. Peer-review©under of the Conference Program B.V. Chairs. 1877-0509 2017responsibility The Authors. Published by Elsevier Peer-review under responsibility of the Conference Program Chairs. 10.1016/j.procs.2017.06.134
136 2
Ahmad Bennakhi et al. / Procedia Computer Science 110 (2017) 135–142 A. Bennakhi et al. / Procedia Computer Science 00 (2016) 000–000
drowsiness, drunkenness, or an eminent epileptic seizure. Implementing this would require no safety procedure since the worst case scenario in all of that is just a misplaced suggestion. There exists an enormous amount of studies that map every state of mind to its corresponding alterations in brain wave signals, but never has there been one to question the application of introducing brain waves to ambient intelligence presence inside cars instead of only controlling the car itself. Sensing brainwaves can also have a tremendous boost to the accuracy contextual navigation and, if perfected, can be an AI present in cars that will make cars a place where interface and user can interact tunefully. Section 1 will be the introduction, while section 2 will discuss studies related to this paper. Section 3 will give the reader some background knowledge about both the topics of brainwaves and the state of mind corresponding to it. Moving on to section 4, we will discuss the details of the case study that this paper is going to tackle. We also will display and analyze the results of the case study with all of its implications in section 4 too. A design of the proposed system that integrates the EEG device with the car is proposed in section 5 with all of its potential applications. The study finally concludes in section 6 with a concise summary of our important findings. Nomenclature Delta waves Theta waves Alpha waves Beta waves Gamma waves
Between 1 Hz and 4 Hz Between 4 Hz and 8 Hz Between 7.5 Hz and 13 Hz Between 13 Hz and 30 Hz Between 30 Hz and 44 Hz
2. Related Work A previous study 4 went through the idea of predicting distractions, using a system that the researchers have programmed to detect. What we plan to do differently from the previously mentioned study is to establish the foundations of a mood sensing system that would not only sense the mood, but also the various mental conditions that could affect the driver using real life driving data (the mentioned study used a computer driving simulator to develop and test their system). Truck drivers were subject of continuous research when it came to driving and EEG. Most of them focused on fatigue/sleepiness during long distance driving 5 . While this paper is old (1993), it marked the beginning of research that focuses on the brain waves of drivers during their journey. More papers emerged with astounding results when it came to predicting the driver’s drowsiness/fatigue 6 7 . All of the previously mentioned studies were done using simulated driving rather than real life testing. A survey was done archiving and comparing all of the studies concerned the detection of drowniness and fatigue while driving, whether it was done using EEG or by other means 8 . It is certainly a good read if someone wants to be informed of the research that was done on the field before 2009. A study 9 had also been conducted to link brain cognition to the mood of individuals. The study also uses an EEG sensor to verify their hypothesis and record the correlation. Predicting the mood while walking was predicated in a study 10 that used electromyogram(EMG) sensors all over the body. Our method is not as intrusive, as it only requires the user to wear a comfortable headband. The use of EEGs to predict alertness and mood was present in a study 11 that wanted to detect the effect of colors on brain waves and concentration. The experiment was not performed on a driver though, since its scope was not specified for that. 3. EEG Brain waves Brain waves have been a under research since the beginning of the 20th century, and they have been thoroughly studied until our present day. A lot of the work has been done on brain waves that researchers can pinpoint which wave lengths correspond with each state of mind. Moreover, the effects of numerous psychiatric medications have been known to produce anticipated brain wave abnormalities. Today’s market offers us a wide range of EEG sensors
Ahmad Bennakhi et al. / Procedia Computer Science 110 (2017) 135–142 A. Bennakhi et al. / Procedia Computer Science 00 (2016) 000–000
137 3
that impose as little discomfort as wearing glasses. As this technology becomes more ubiquitous we will examine the potential applications of the technology while driving a vehicle. The brain waves classification is a nomenclature, especially when it comes to the boundary of each band. Our EEG headband can detect 5 bands, therefore we based our classification of the EEG spectrum on the headband that was used in our study and listed in the nomenclature. The following brain waves are listed according to what the majority of sources consider them to be. The delta band is the lowest one of these wave bands (below 4 Hz). It is mainly associated with deep dreamless sleep, but is also present when a person is doing tasks that require continuous attention for a long period of time 12 . Theta waves (4 Hz to 8 Hz) on the other hand is present in REM sleep, meditation, drowsiness, idling, and repression of a certain action or response 12 . The Alpha (8 Hz to 13 Hz) band is related to relaxation, daydreaming, and closed eyes. In an activity such as watching TV, alpha waves are often the dominant band. Beta waves (13 Hz to 30 Hz) signify alertness, focus, and anxiousness. Beta waves could range from a calm active state of mind to intense or stress to mild obsessiveness. The last band on the list is Gamma (30+ Hz), which is associated with hyper alertness and high learning. Gamma waves are evident during cross-modal sensory processing and when a person is thinking very deeply looking for insights 13 . The following points can be summarized in table 1 (shown below). Table 1. EEG bands and the state of mind that are associated with
Band
Frequency (t)
Associated state of mind (t)
Delta Theta Alpha Beta Gamma
Less than 4 Hz 4 Hz to 8 Hz 8 Hz to 15 Hz 16 Hz to 31 Hz More than 31 Hz
Deep dreamless sleep or K-complex 14 15 REM, meditation 16 , drowsiness 17 , idling or repression of action/response 18 Daydreaming, relaxed, being awake with eyes closed 15 Awake, alert, engaged, focused 19 , or anxious 20 Hyper alert and cross-modal sensory processing (perception that combines two different senses, such as sound and sight) 21
Some books have even described personality types that have tendencies to dominate certain bands 22 . With all of that information flowing, we are going to talk about the technicality of harvesting it in a particular case study and applying it in a useful way. 4. Case Study: Mood Detection while Driving 4.1. Method The brain wave headset that we used in this case study is the Muse™ headband. Muse™ has 7 sensors that are strategically placed on the forehead, temple, and behind the ears. We kept in mind that the headset has to be comfortable and reliable at the same time, that is why we chose Muse™ in particular. The headband also has an adequate library that makes coding a custom application much easier. The experiment begins with opening a custom mobile application that was created specifically for this study. This mobile application syncs with the headset placed on the volunteer’s head. After a connection is made between the headband and the mobile application, the session begins with a simple question of “How are you feeling?” with 7 options to choose from. After doing so, the volunteer drives along his intended route in a regular fashion without being distracted by anything related to the experiment. Once the volunteer reaches his destination, he would end the session and answer three more questions. The first being “How are you feeling?” with the reply options that were available earlier on, and the second question is to determine the level of traffic by selecting from 3 choices. The last question will ask if the user telephoned anyone during his trip. All of the test subjects are over 18 and live in Kuwait City, which is considered a big metropolitan city. All sessions recorded are over 5 minutes long and have adequate signal strength from the headset. The application that is being used is installed in smartphones that are given to the volunteers. A total of 11 volunteers(6 males and 5 females)
Ahmad Bennakhi et al. / Procedia Computer Science 110 (2017) 135–142
138 4
A. Bennakhi et al. / Procedia Computer Science 00 (2016) 000–000
Figure 1. A participant using the EEG device
participated in this experiment and each were asked to perform a total of 5 sessions. It is worth mentioning that some sessions were discarded due to the poor quality and device misuse, cumulatively the total number of sessions is 44. The sessions were all recorded by the Muse™ headband, and the parameter that was used in our study was the absolute band power. The absolute band power for a given frequency range is the logarithm of the sum of the Power Spectral Density of the EEG data over that frequency range. They are provided for each of the four to six channels/electrode sites on Muse 23 . The maximum power recorded out of all of the six electrodes was taken as the most valid value, since it is the strongest signal detected from the brain at the specific frequency band. The data coming from the headband would be flagged in the following way: • If the mood in the beginning matches the mood in the end of his/her ride then the whole session would be flagged in the mood specified by the driver as it is assumed that there were no major mood changes through their journey. • If the mood recorded in the beginning does not match the mood in the end, then the first 1000 (the average interval between transfers is 0.1 seconds) transfers of brain wave data are flagged with the mood recorded in the beginning of the session. Similarly, the last 1000 transfers are flagged with the mood recorded at the end of the session.
4.2. Results The sessions were analyzed using the bivariate correlation method: Kendall’s tau-b. The data was first separated into different files according to mood then analyzed separately for correlations. The variables that were analyzed were the brain wave bands: Alpha, Beta, Delta, Theta, and Gamma. The result was 7 correlation tables(each mood had a table). The combined results of these tables can be seen in table 2. The correlations that were significantly positive are listed beside a plus sign, while the negative ones are listed beside a negative dash sign. The results that yielded correlations that were close to zero are listed beside bullet points. The Alpha-Beta correlation coefficient is high (more than doubled) for non-negative emotions such as neutrality, happiness, and tired. While sadness shows a bit of a significant inverse correlation. Moreover, sadness varies from the other emotions in the case that it is the only emotion with a relatively high Alpha-Delta correlation. The AlphaTheta band also has relatively high correlation coefficients when it comes to emotions such as hunger and sadness. The Alpha-Gamma correlation is unique since it has a relatively high inverse correlation coefficient when it comes to happiness and sadness, which are an unlikely pair even when compared with the other band correlations in this study. The Alpha-Gamma correlation is highest on the positive side when people reported feeling “neutral”, which may suggest that emotions whether happy or sad contribute to a negative correlation coefficient, and lack of them contribute to the positive end. Negative emotions such as anger, stress, hunger, and sadness all have relatively high inverse Beta-Delta correlation coefficients especially anger. Beta also has a statistically significant inverse correlation with Theta when people recorded being stressed and angry. Across all emotions, the Beta-Gamma correlation coefficients are highly positive with somber moods such as being tired and sad, ranking the lowest. Higher positive correlation coefficients are also present between the Delta-Theta bands with stress and being tired, having the highest of them all. The list of negative emotions (hunger, anger, and sadness) show common features once again by having similar negative correlation
Ahmad Bennakhi et al. / Procedia Computer Science 110 (2017) 135–142 A. Bennakhi et al. / Procedia Computer Science 00 (2016) 000–000
139 5
Table 2. The correlation coefficient is written inside the brackets of each mood. Highlighted results are deemed to be of interest.
Alpha
Beta +Neutral(0.308), Tired (0.308), Happy ( 0.3) Hungry( 0.142), Stressed(0.109) •Angry( -0.068) - Sad( -0.132)
Beta
Delta
Theta
Delta +Sad(0.248) •Neutral(0.082), Tired(0.072), Hungry(0.046), Angry(0.019), Happy(0.013) , Stressed(-0.054) +Happy(0.077) •Tired(0.018), Neutral(-0.023), -Stressed(-0.111), Hungry(-0.113), Sad(-0.181), Angry(-0.242)
Theta +Sad(0.232), Hungry(0.198), Tired(0.144) • Happy(0.088), Neutral(0.076), Angry(0.06), Stressed(-0.006) +Sad(0.14), Tired(0.109) • Hungry(0.018), Happy(-0.008), Neutral(-0.022) -Stressed(-0.165), Angry(-0.197) +Stressed(0.395), Tired(0.319), Hungry(0.266), Happy(0.240), Angry(0.234), Neutral(0.184), •Sad(0.058)
Gamma +Neutral(0.165), Tired(0.096), Hungry(0.084), Stressed(0.064) •Angry(-0.002) -Sad(-0.268), Happy(-0.28) +Hungry(0.605), Stressed(0.589), Neutral(0.535), Happy(0.517), Angry(0.501), Tired(0.495), Sad(0.317) • Happy(0.018), Tired(0.014), Neutral(-0.04), Stressed(-0.066) -Hungry(-0.165), Angry(-0.192), Sad(-0.242) +Tired(0.111) •Sad(0.081), Neutral(-0.019), Happy(-0.022), Hungry(-0.061) -Stressed(-0.172), Angry(-0.312)
coefficients between the Delta-Gamma bands. Finally, the Theta-Gamma correlation coefficients show an intense anger mood on the significant negative end and a worn-out tired mood on the positive end of the correlation table. In between several brain wave bands, we can notice a set of emotions that mostly go together. Most noticeably sadness and hunger tend to have similar correlations. Less negative moods such as being happy, tired, and neutral have also noted to stick together. That being said, there are several cross band correlations that can act as indicators to point out the mood that the driver is feeling. For example, analyzing the Alpha-Gamma band can reveal if the person is neutral or happy, both of which normally stick together but differ greatly in this case. 4.3. Implications This is just a preliminary proof of concept study and it already produced results that could be used in real life applications. Only one type of correlation method was applied to the data, so further analysis would no doubt reveal more interesting data. A car interface application that monitors and compares bandwidths would definitely be able to differentiate between moods given that the subject is researched more thoroughly with more participants. Moreover, if the car application is programmed to adjust to the driver’s personal brain wave patterns, then an even more concise mood recognition system could be developed. Mood recognition using EEG is a fairly new area that is being explored. So if this system can also be implemented in autonomous cars then an array of new possibilities will open up for new applications that can utilize an EEG headband. The user of such a system could be monitored not just for mood, but other conditions as well such as
Ahmad Bennakhi et al. / Procedia Computer Science 110 (2017) 135–142 A. Bennakhi et al. / Procedia Computer Science 00 (2016) 000–000
140 6
epileptic seizures, fatigue or alcohol abuse 24 . These areas are drenched with already existing research studies that would help build a basis for a system that would detect these conditions. 5. Proposed System Design Before we begin to outline how the system works, here are the guidelines that will be the basis of our proposed system: 1. The system will not commit to any action unless the user gives his/her consent (if the driver has the capability of making decisions). 2. The system should not have any control over the car’s driving system unless autonomous driving becomes approved by the concerned government associations. 3. The system should impose minimal or no distraction at all on the driver. 4. The driver should not have to order the system to do anything (the system has to anticipate his needs), since it is purely based on the system giving suggestions or asking for consent to perform a task. A brain wave sensor has to be used so that the system can be implemented. The brain wave headband can have auxiliary parts that make it have more than one particular function (e.g. glasses, Bluetooth headset, or a visor). As seen in the figure below, the sensor that we used in our case study can be an integral part of many accessories. Therefore, it would be easy for users to wear it casually and still reap the benefits of the implemented system. There are two flavors of Homonoia: Table 3. The flavors of Homonoia
Flavor
Can Access
Requirements
Skinny
Car media player, Smartphone, Car Interface, and GPS
An integrated application that is installed in the car’s interface system
Autonomous
In addition to the skinny flavor control privileges, it also has control over the car seat and driving system
In addition to the skinny flavor requirements, it also requires a fully secure autonomous driving system that would authorize emergency intervention when needed.
5.1. Skinny Self-driving cars are in no way fail-proof. A report 25 compiled by the DMV summarized the severity of the situation by documenting the total of times that self-driving cars disengaged, whether the cause of the disengagement was a hardware/software malfunction or just the driver sensing that something went wrong. That is why we need an implementation that would isolate the car controls from the Homonoia system. The skinny flavor would only have control over the car’s interface and smartphone only. It should also be clear that smartphone access is purely optional, but will also make the integration of the system much more comprehensive and efficient. The skinny Homonoia flavor has a unilateral sequence of action: Recognition: While the headset is forwarding EEG data to the car’s interface system, there is a constant search for parameters to recognize the mood (e.g. angry, happy...etc.) . Interaction: A simple relevant question is directed towards the user so that an action can be commenced (e.g. “Do you want me to turn on the soothing music?” Or “Theres a caf nearby, do you want to grab a cup of coffee from there?”). The user would respond by a yes, no, or another reaction that would be recognized by the natural language voice recognition system.
Ahmad Bennakhi et al. / Procedia Computer Science 110 (2017) 135–142 A. Bennakhi et al. / Procedia Computer Science 00 (2016) 000–000
141 7
Figure 2. The proposed system summary
Recording: All replies from the user is documented as well as the time and other variables that could affect the circumstances, so that future suggestions could become adaptively better. The system is bound to get better with each suggestion, so that the number of hits per suggestion would steadily increase, eventually reaching homonoia status. 5.2. Autonomous In addition to all of the functions that the skinny flavor offers, the Autonomous flavor offers an emergency response option. There are numerous studies that demonstrated the ability to detect mental conditions such as encephalopathy, strokes, sleep disorders, disorientation, alcohol abuse, epilepsy, and many other brain abnormalities 24 . The profile for each brain condition can be saved and referenced so that Homonoia checks for its occurrence. This flavor can be implemented when self-driving cars reach a sufficient level of safety. In case a brain wave abnormality is detected, the autonomous driving system would be placed on alert mode. Being on alert mode still keeps the driver in control of the car, but stays on the look for driving hazards such as collisions or swaying out of the road. Furthermore, a timed enquiry will be posted on the cars interface to make sure if the driver is in good mental condition. If a user manages to respond with an appropriate answer, only then that car would return to normal mode. Otherwise, if the driver did not respond, or a dangerous driving pattern was detected the car would transition to panic mode. The vehicle would have to either park on the side of the road or drive autonomously to the nearest hospital. The driver’s seat would also be adjusted to match the condition suitable for the user’s brain abnormality. 6. Conclusion We have exhibited how brain waves can act as indicators of numerous brain states. The previous studies that integrated brain waves with technology have focused mainly on it having an active role when driving, which needs a
142 8
Ahmad Bennakhi et al. / Procedia Computer Science 110 (2017) 135–142 A. Bennakhi et al. / Procedia Computer Science 00 (2016) 000–000
lot of precaution and may even be rendered obsolete with the emergence of autonomous cars. Therefore, we decided to examine the passive role that brain waves can take to compliment other technologies rather than compete with them. We have carried out an experiment that could demonstrate whether we could harness certain correlation patterns of brain wave bands to pin point the driver’s mood. The results show that certain negative moods usually group together with similar correlation coefficients between several brain wave bands. We have also proposed a system that acts in a way that adds minimal workload on the user while integrating him/her as much as possible with an intelligent car interface system. Of course, more research needs to be done in this area in order for this system to be applied to reality in an efficient way. Acknowledgements Special thanks go to Ali Al-Hasan for providing us with a number of volunteers willing to take part in the experiment. References 1. West, W.C.. Hellenic homonoia and the new decree from plataea. Greek, Roman and Byzantine Studies 1977;18(4):307. 2. Chinese researchers unveil brain powered car. 2015. URL: http://www.reuters.com/article/ us-china-brainpower-car-idUSKBN0TQ23620151207. 3. Braindriver. 2015. URL: http://autonomos-labs.com/projects/past-projects/braindriver/. 4. Wang, S., Zhang, Y., Wu, C., Darvas, F., Chaovalitwongse, W.A.. Online prediction of driver distraction based on brain activity patterns. IEEE Transactions on Intelligent Transportation Systems 2015;16(1):136–150. 5. Kecklund, G., Åkerstedt, T.. Sleepiness in long distance truck driving: an ambulatory eeg study of night driving. Ergonomics 1993;36(9):1007–1017. 6. Lal, S.K., Craig, A., Boord, P., Kirkup, L., Nguyen, H.. Development of an algorithm for an eeg-based driver fatigue countermeasure. Journal of safety Research 2003;34(3):321–328. 7. Kim, J.Y., Jeong, C.H., Jung, M.J., Park, J.H., Jung, D.H.. Highly reliable driving workload analysis using driver electroencephalogram (eeg) activities during driving. International Journal of Automotive Technology 2013;14(6):965–970. 8. Liu, C.C., Hosking, S.G., Lenn´e, M.G.. Predicting driver drowsiness using vehicle measures: Recent insights and future challenges. Journal of safety research 2009;40(4):239–245. 9. Yuan, J., Xu, S., Yang, J., Liu, Q., Chen, A., Zhu, L., et al. Pleasant mood intensifies brain processing of cognitive control: Erp correlates. Biological psychology 2011;87(1):17–24. 10. Kinase, Y., Venture, G.. Estimating mood variation from mpf of emg during walking. In: Engineering in Medicine and Biology Society (EMBC), 2013 35th Annual International Conference of the IEEE. IEEE; 2013, p. 4981–4984. 11. Plitnick, B., Figueiro, M., Wood, B., Rea, M.. The effects of red and blue light on alertness and mood at night. Lighting Research & Technology 2010;42(4):449–458. 12. Kirmizi-Alsan, E., Bayraktaroglu, Z., Gurvit, H., Keskin, Y.H., Emre, M., Demiralp, T.. Comparative analysis of event-related potentials during go/nogo and cpt: decomposition of electrophysiological markers of response inhibition and sustained attention. Brain research 2006;1104(1):114–128. 13. Kisley, M.A., Cornwell, Z.M.. Gamma and beta neural activity evoked during a sensory gating paradigm: effects of auditory, somatosensory and cross-modal stimulation. Clinical neurophysiology 2006;117(11):2549–2563. 14. De Gennaro, L., Ferrara, M., Bertini, M.. The spontaneous k-complex during stage 2 sleep: is it the forerunnerof delta waves? Neuroscience letters 2000;291(1):41–43. 15. Marcuse, L.V., Fields, M.C., Yoo, J.J.. Rowan’s Primer of EEG. Elsevier Health Sciences; 2015. 16. Kubota, Y., Sato, W., Toichi, M., Murai, T., Okada, T., Hayashi, A., et al. Frontal midline theta rhythm is correlated with cardiac autonomic activities during the performance of an attention demanding meditation procedure. Cognitive Brain Research 2001;11(2):281–287. 17. Makeig, S., Jung, T.P.. Tonic, phasic, and transient eeg correlates of auditory awareness in drowsiness. Cognitive Brain Research 1996;4(1):15–25. 18. Di Lazzaro, V., Pilato, F., Saturno, E., Oliviero, A., Dileone, M., Mazzone, P., et al. Theta-burst repetitive transcranial magnetic stimulation suppresses specific excitatory circuits in the human motor cortex. The Journal of physiology 2005;565(3):945–950. 19. Dustman, R.E., Boswell, R.S., Porter, P.B.. Beta brain waves as an index of alertness. Science 1962;137(3529):533–534. 20. Hammond, D.C.. Neurofeedback with anxiety and affective disorders. Child and adolescent psychiatric clinics of North America 2005;14(1):105–123. 21. Jemmer, P.. Getting in a (brain-wave) state through entrainment, meditation and hypnosis. Hypnotherapy Journal 2009;2:24–29. 22. Kropotov, J.D.. Quantitative EEG, event-related potentials and neurotherapy. Academic Press; 2010. 23. Available data. 2017. URL: http://developer.choosemuse.com/research-tools/available-data#Absolute_Band_Powers. 24. Eeg: Medlineplus medical encyclopedia. 2017. URL: https://www.nlm.nih.gov/medlineplus/ency/article/003931.htm. 25. Vehicles, C.D.o.M.. Autonomous vehicle disengagement reports 2016. 2016. URL: https://www.dmv.ca.gov/portal/dmv/detail/ vr/autonomous/disengagement_report_2016.