The last few decades have seen widespread advances in technological means to characterise observable aspects of human behaviour such as gaze or posture. Among others, these developments have also led to significant advances in social robotics. At the same time, however, social robots are still largely evaluated in idealised or laboratory conditions, and it remains unclear whether the technological progress is sufficient to let such robots move “into the wild”. In this paper, we characterise the problems that a social robot in the real world may face, and review the technological state of the art in terms of addressing these. We do this by considering what it would entail to automate the diagnosis of Autism Spectrum Disorder (ASD). Just as for social robotics, ASD diagnosis fundamentally requires the ability to characterise human behaviour from observable aspects. However, therapists provide clear criteria regarding what to look for. As such, ASD diagnosis is a situation that is both relevant to real-world social robotics and comes with clear metrics. Overall, we demonstrate that even with relatively clear therapist-provided criteria and current technological progress, the need to interpret covert behaviour cannot yet be fully addressed. Our discussions have clear implications for ASD diagnosis, but also for social robotics more generally. For ASD diagnosis, we provide a classification of criteria based on whether or not they depend on covert information and highlight present-day possibilities for supporting therapists in diagnosis through technological means. For social robotics, we highlight the fundamental role of covert behaviour, show that the current state-of-the-art is unable to charact
In recent years, the field of Human-Robot Interaction (HRI) has seen an increasingdemand for technologies that can recognize and adapt to human behaviors and internalstates (e.g., emotions and intentions). Psychological research suggests that humanmovements are important for inferring internal states. There is, however, a need to betterunderstand what kind of information can be extracted from movement data, particularlyin unconstrained, natural interactions. The present study examines which internal statesand social constructs humans identify from movement in naturalistic social interactions.Participants either viewed clips of the full scene or processed versions of it displaying2D positional data. Then, they were asked to fill out questionnaires assessing their socialperception of the viewed material. We analyzed whether the full scene clips were moreinformative than the 2D positional data clips. First, we calculated the inter-rater agreementbetween participants in both conditions. Then, we employed machine learning classifiersto predict the internal states of the individuals in the videos based on the ratingsobtained. Although we found a higher inter-rater agreement for full scenes comparedto positional data, the level of agreement in the latter case was still above chance,thus demonstrating that the internal states and social constructs under study wereidentifiable in both conditions. A factor analysis run on participants’ responses showedthat participants identified the constructs interaction imbalance, interaction valence andengagement regardless of video condition. The machine learning classifiers achieveda similar performance in both conditions, again supporting the idea that movementalone carries relevant information. Overall, our results suggest it is reasonable to expecta machine learning algorithm, and consequently a robot, to successfully decode andclassify a range of internal states and social constructs using low-dimensional data (suchas the movements and poses of observed individuals) as input.
We present a dataset of behavioral data recorded from 61 children diagnosed with Autism Spectrum Disorder (ASD). The data was collected during a large-scale evaluation of Robot Enhanced Therapy (RET). The dataset covers over 3000 therapy sessions and more than 300 hours of therapy. Half of the children interacted with the social robot NAO supervised by a therapist. The other half, constituting a control group, interacted directly with a therapist. Both groups followed the Applied Behavior Analysis (ABA) protocol. Each session was recorded with three RGB cameras and two RGBD (Kinect) cameras, providing detailed information of children’s behavior during therapy. This public release of the dataset comprises body motion, head position and orientation, and eye gaze variables, all specified as 3D data in a joint frame of reference. In addition, metadata including participant age, gender, and autism diagnosis (ADOS) variables are included. We release this data with the hope of supporting further data-driven studies towards improved therapy methods as well as a better understanding of ASD in general.
It is evident that recently reported robot-assisted therapy systems for assessment of children with autism spectrum disorder (ASD) lack autonomous interaction abilities and require significant human resources. This paper proposes a sensing system that automatically extracts and fuses sensory features such as body motion features, facial expressions, and gaze features, further assessing the children behaviours by mapping them to therapist-specified behavioural classes. Experimental results show that the developed system has a capability of interpreting characteristic data of children with ASD, thus has the potential to increase the autonomy of robots under the supervision of a therapist and enhance the quality of the digital description of children with ASD. The research outcomes pave the way to a feasible machine-assisted system for their behaviour assessment. IEEE
A growing body of evidence in cognitive science and neuroscience points towards the existence of a deep interconnection between cognition, perception and action. According to this embodied perspective language is grounded in the sensorimotor system and language understanding is based on a mental simulation process (Jeannerod, 2007; Gallese, 2008; Barsalou, 2009). This means that during action words and sentence comprehension the same perception, action, and emotion mechanisms implied during interaction with objects are recruited. Among the neural underpinnings of this simulation process an important role is played by a sensorimotor matching system known as the mirror neuron system (Rizzolatti and Craighero, 2004). Despite a growing number of studies, the precise dynamics underlying the relation between language and action are not yet well understood. In fact, experimental studies are not always coherent as some report that language processing interferes with action execution while others find facilitation. In this work we present a detailed neural network model capable of reproducing experimentally observed influences of the processing of action-related sentences on the execution of motor sequences. The proposed model is based on three main points. The first is that the processing of action-related sentences causes the resonance of motor and mirror neurons encoding the corresponding actions. The second is that there exists a varying degree of crosstalk between neuronal populations depending on whether they encode the same motor act, the same effector or the same action-goal. The third is the fact that neuronal populations’ internal dynamics, which results from the combination of multiple processes taking place at different time scales, can facilitate or interfere with successive activations of the same or of partially overlapping pools.
Automated driving needs unprecedented levels of reliably and safety before marked deployment. The average human driver fatal accident rate is 1 every 100 million miles. Automated vehicles will have to provably best these figures. This paper introduces the notion of dream-like mechanisms as a simulation technology to produce a large number of hypothetical design and test scenarios - especially focusing on variations of more frequent dangerous and near miss events. Grounded in the simulation hypothesis of cognition, we show here some principles for effective simulation mechanisms and an artificial cognitive system architecture that can learn from the simulated situations.
Engagement is essential to meaningful social interaction between humans. Understanding the mechanisms by which we detect engagement of other humans can help us understand how we can build robots that interact socially with humans. However, there is currently a lack of measurable engagement constructs on which to build an artificial system that can reliably support social interaction between humans and robots. This paper proposes a definition, based on motivation theories, and outlines a framework to explore the idea that engagement can be seen as specific behaviors and their attached magnitude or intensity. This is done by the use of data from multiple sources such as observer ratings, kinematic data, audio and outcomes of interactions. We use the domain of human-robot interaction in order to illustrate the application of this approach. The framework further suggests a method to gather and aggregate this data. If certain behaviors and their attached intensities co-occur with various levels of judged engagement, then engagement could be assessed by this framework consequently making it accessible to a robotic platform. This framework could improve the social capabilities of interactive agents by adding the ability to notice when and why an agent becomes disengaged, thereby providing the interactive agent with an ability to reengage him or her. We illustrate and propose validation of our framework with an example from robot-assisted therapy for children with autism spectrum disorder. The framework also represents a general approach that can be applied to other social interactive settings between humans and robots, such as interactions with elderly people.
Behavioral studies on the activation of affordances by understanding observation and action sentences on graspable objects show a direct relationship between the canonical orientation of graspable objects, their dimension and the kind of grip required by those objects to be grasped. The present work introduces the concepts of Dynamic Field Theory for modeling the results observed in the behavioral studies previously mentioned. The model was not only able to replicate qualitatively similar results regarding reaction times, but also the identification of same versus different object and the distinction between observable versus action sentences. The model shows the potential of dynamic field theory for the design and implementation of brain inspired cognitive systems. © 2012 Springer-Verlag.
Robot-Assisted Therapy (RAT) has successfully been used to improve social skills in children with autism spectrum disorders (ASD) through remote control of the robot in so-called Wizard of Oz (WoZ) paradigms.However, there is a need to increase the autonomy of the robot both to lighten the burden on human therapists (who have to remain in control and, importantly, supervise the robot) and to provide a consistent therapeutic experience. This paper seeks to provide insight into increasing the autonomy level of social robots in therapy to move beyond WoZ. With the final aim of improved human-human social interaction for the children, this multidisciplinary research seeks to facilitate the use of social robots as tools in clinical situations by addressing the challenge of increasing robot autonomy.We introduce the clinical framework in which the developments are tested, alongside initial data obtained from patients in a first phase of the project using a WoZ set-up mimicking the targeted supervised-autonomy behaviour. We further describe the implemented system architecture capable of providing the robot with supervised autonomy.
The purpose of the present experiment is to further understand the effect of levels of processing (top-down vs. bottom-up) on the perception of movement kinematics and primitives for grasping actions in order to gain insight into possible primitives used by the mirror system. In the present study, we investigated the potential of identifying such primitives using an action segmentation task. Specifically, we investigated whether or not segmentation was driven primarily by the kinematics of the action, as opposed to high-level top-down information about the action and the object used in the action. Participants in the experiment were shown 12 point-light movies of object-centered hand/arm actions that were either presented in their canonical orientation together with the object in question (top-down condition) or upside down (inverted) without information about the object (bottom-up condition). The results show that (1) despite impaired high-level action recognition for the inverted actions participants were able to reliably segment the actions according to lower-level kinematic variables, (2) segmentation behavior in both groups was significantly related to the kinematic variables of change in direction, velocity, and acceleration of the wrist (thumb and finger tips) for most of the included actions. This indicates that top-down activation of an action representation leads to similar segmentation behavior for hand/arm actions compared to bottom-up, or local, visual processing when performing a fairly unconstrained segmentation task. Motor primitives as parts of more complex actions may therefore be reliably derived through visual segmentation based on movement kinematics.
The affective motion of humans conveys messages that other humans perceive and understand without conventional linguistic processing. This ability to classify human movement into meaningful gestures or segments plays also a critical role in creating social interaction between humans and robots. In the research presented here, grasping and social gesture recognition by humans and four machine learning techniques (k-Nearest Neighbor, Locality-Sensitive Hashing Forest, Random Forest and Support Vector Machine) is assessed by using human classification data as a reference for evaluating the classification performance of machine learning techniques for thirty hand/arm gestures. The gestures are rated according to the extent of grasping motion on one task and the extent to which the same gestures are perceived as social according to another task. The results indicate that humans clearly rate differently according to the two different tasks. The machine learning techniques provide a similar classification of the actions according to grasping kinematics and social quality. Furthermore, there is a strong association between gesture kinematics and judgments of grasping and the social quality of the hand/arm gestures. Our results support previous research on intention-from-movement understanding that demonstrates the reliance on kinematic information for perceiving the social aspects and intentions in different grasping actions as well as communicative point-light actions.
The neural circuits that control grasping and perform related visual processing have been studied extensively in macaque monkeys. We are developing a computational model of this system, in order to better understand its function, and to explore applications to robotics. We recently modelled the neural representation of three-dimensional object shapes, and are currently extending the model to produce hand postures so that it can be tested on a robot. To train the extended model, we are developing a large database of object shapes and corresponding feasible grasps. Finally, further extensions are needed to account for the influence of higher-level goals on hand posture. This is essential because often the same object must be grasped in different ways for different purposes. The present paper focuses on a method of incorporating such higher-level goals. A proof-of-concept exhibits several important behaviours, such as choosing from multiple approaches to the same goal. Finally, we discuss a neural representation of objects that supports fast searching for analogous objects.
Rapid technical advancements have led to dramatically improved abilities for artificial agents, and thus opened up for new ways of cooperation between humans and them, from disembodied agents such as Siris to virtual avatars, robot companions, and autonomous vehicles. It is therefore relevant to study not only how to maintain appropriate cooperation, but also where the responsibility for this resides and/or may be affected. While there are previous organisations and categorisations of agents and HAI research into taxonomies, situations with highly responsible artificial agents are rarely covered. Here, we propose a way to categorise agents in terms of such responsibility and agent autonomy, which covers the range of cooperation from humans getting help from agents to humans providing help for the agents. In the resulting diagram presented in this paper, it is possible to relate different kinds of agents with other taxonomies and typical properties. A particular advantage of this taxonomy is that it highlights under what conditions certain effects known to modulate the relationship between agents (such as the protégé effect or the "we"-feeling) arise.
Human-robot interaction (HRI) is fundamentally concerned with studying the interaction between humans and robots. While it is still a relatively young field, it can draw inspiration from other disciplines studying human interaction with other types of agents. Often, such inspiration is sought from the study of human-computer interaction (HCI) and the social sciences studying human-human interaction (HHI). More rarely, the field also turns to human-animal interaction (HAI).
In this paper, we identify two distinct underlying motivations for making such comparisons: to form a target to recreate orto obtain a benchmark (or baseline) for evaluation. We further highlight relevant (existing) overlap between HRI and HAI, and identify specific themes that are of particular interest for further trans-disciplinary exploration. At the same time, since robots and animals are clearly not the same, we also discuss important differences between HRI and HAI, their complementarity notwithstanding. The overall purpose of this discussion is thus to create an awareness of the potential mutual benefit between the two disciplines and to describe opportunities that exist for future work, both in terms of new domains to explore, and existing results to learn from.
The traffic domain is increasingly inhabited by vehicles with driving support systems and automation to the degree where the idea of fully autonomous vehicles is gaining popularity as a credible prediction about the near future. As more aspects of driving become automated, the role of the driver, and the way they perceive their vehicle, surroundings, and fellow road users, change. To address some of the emerging kinds of interaction between different agents in the traffic environment, it is important to take social phenomena and abilities into account, even to the extent of considering highly automated vehicles to be social agents in their own right. To benefit from that, it is important to frame the perception of the traffic environment, as well as the road users in it, in an appropriate theoretical context. We propose that there are helpful concepts related to functional and subjective perception, derived from gestalt psychology and Umweltlehre, that can fill this theoretical need, and support better understanding of vehicles of various degrees of automation.
The term ‘multimodality’ has come to take on several somewhat different meanings depending on the underlying theoretical paradigms and traditions, and the purpose and context of use. The term is closely related to embodiment, which in turn is also used in several different ways. In this paper, we elaborate on this connection and propose that a pragmatic and pluralistic stance is appropriate for multimodality. We further propose a distinction between first and second order effects of multimodality; what is achieved by multiple modalities in isolation and the opportunities that emerge when several modalities are entangled. This highlights questions regarding ways to cluster or interchange different modalities, for example through redundancy or degeneracy. Apart from discussing multimodality with respect to an individual agent, we further look to more distributed agents and situations where social aspects become relevant.
In robotics, understanding the various uses and interpretations of these terms can prevent miscommunication when designing robots, as well as increase awareness of the underlying theoretical concepts. Given the complexity of the different ways in which multimodality is relevant in social robotics, this can provide the basis for negotiating appropriate meanings of the term at a case by case basis.
Learning from only real-world collected data can be unrealistic and time consuming in many scenario. One alternative is to use synthetic data as learning environments to learn rare situations and replay buffers to speed up the learning. In this work, we examine the hypothesis of how the creation of the environment affects the training of reinforcement learning agent through auto-generated environment mechanisms. We take the autonomous vehicle as an application. We compare the effect of two approaches to generate training data for artificial cognitive agents. We consider the added value of curriculum learning—just as in human learning—as a way to structure novel training data that the agent has not seen before as well as that of using a replay buffer to train further on data the agent has seen before. In other words, the focus of this paper is on characteristics of the training data rather than on learning algorithms. We therefore use two tasks that are commonly trained early on in autonomous vehicle research: lane keeping and pedestrian avoidance. Our main results show that curriculum learning indeed offers an additional benefit over a vanilla reinforcement learning approach (using Deep-Q Learning), but the replay buffer actually has a detrimental effect in most (but not all) combinations of data generation approaches we considered here. The benefit of curriculum learning does depend on the existence of a well-defined difficulty metric with which various training scenarios can be ordered. In the lane-keeping task, we can define it as a function of the curvature of the road, in which the steeper and more occurring curves on the road, the more difficult it gets. Defining such a difficulty metric in other scenarios is not always trivial. In general, the results of this paper emphasize both the importance of considering data characterization, such as curriculum learning, and the importance of defining an appropriate metric for the task.
Self-driving cars not only solve the problem of navigating safely from location A to location B; they also have to deal with an abundance of (sometimes unpredictable) factors, such as traffic rules, weather conditions, and interactions with humans. Over the last decades, different approaches have been proposed to design intelligent driving systems for self-driving cars that can deal with an uncontrolled environment. Some of them are derived from computationalist paradigms, formulating mathematical models that define the driving agent, while other approaches take inspiration from biological cognition. However, despite the extensive work in the field of self-driving cars, many open questions remain. Here, we discuss the different approaches for implementing driving systems for self-driving cars, as well as the computational paradigms from which they originate. In doing so, we highlight two key messages: First, further progress in the field might depend on adapting new paradigms as opposed to pushing technical innovations in those currently used. Specifically, we discuss how paradigms from cognitive systems research can be a source of inspiration for further development in modeling driving systems, highlighting emergent approaches as a possible starting point. Second, self-driving cars can themselves be considered cognitive systems in a meaningful sense, and are therefore a relevant, yet underutilised resource in the study of cognitive mechanisms. Overall, we argue for a stronger synergy between the fields of cognitive systems and self-driving vehicles.
The controller of an autonomous vehicle needsthe ability to learn how to act in different driving scenariosthat it may face. A significant challenge is that it is difficult,dangerous, or even impossible to experience and explore variousactions in situations that might be encountered in the realworld. Autonomous vehicle control would therefore benefitfrom a mechanism that allows the safe exploration of actionpossibilities and their consequences, as well as the ability tolearn from experience thus gained to improve driving skills.In this paper we demonstrate a methodology that allows alearning agent to create simulations of possible situations. Thesesimulations can be chained together in a sequence that allowsthe progressive improvement of the agent’s performance suchthat the agent is able to appropriately deal with novel situationsat the end of training. This methodology takes inspirationfrom the human ability to imagine hypothetical situations usingepisodic simulation; we therefore refer to this methodology asepisodic imagination.An interesting question in this respect is what effect thestructuring of such a sequence of episodic imaginations hason performance. Here, we compare a random process to astructured one and initial results indicate that a structuredsequence outperforms a random one.
Almost all animals exploit vocal signals for a range of ecologically-motivated purposes: detecting predators/prey and marking territory, expressing emotions, establishing social relations and sharing information. Whether it is a bird raising an alarm, a whale calling to potential partners, a dog responding to human commands, a parent reading a story with a child, or a business-person accessing stock prices using \emph{Siri}, vocalisation provides a valuable communication channel through which behaviour may be coordinated and controlled, and information may be distributed and acquired. Indeed, the ubiquity of vocal interaction has led to research across an extremely diverse array of fields, from assessing animal welfare, to understanding the precursors of human language, to developing voice-based human-machine interaction. Opportunities for cross-fertilisation between these fields abound; for example, using artificial cognitive agents to investigate contemporary theories of language grounding, using machine learning to analyse different habitats or adding vocal expressivity to the next generation of language-enabled autonomous social agents. However, much of the research is conducted within well-defined disciplinary boundaries, and many fundamental issues remain. This paper attempts to redress the balance by presenting a comparative review of vocal interaction within-and-between humans, animals and artificial agents (such as robots), and it identifies a rich set of open research questions that may benefit from an inter-disciplinary analysis.
According to the simulation hypothesis, mental imagery can be explained in terms of predictive chains of simulated perceptions and actions, i.e., perceptions and actions are reactivated internally by our nervous system to be used in mental imagery and other cognitive phenomena. Our previous research shows that it is possible but not trivial to develop simulations in robots based on the simulation hypothesis. While there are several previous approaches to modelling mental imagery and related cognitive abilities, the origin of such internal simulations has hardly been addressed. The inception of simulation (InSim) hypothesis suggests that dreaming has a function in the development of simulations by forming associations between experienced, non-experienced but realistic, and even unrealistic perceptions. Here, we therefore develop an experimental set-up based on a simple simulated robot to test whether such dream-like mechanisms can be used to instruct research into the development of simulations and mental imagery-like abilities. Specifically, the hypothesis is that dreams' informing the construction of simulations lead to faster development of good simulations during waking behaviour. The paper presents initial results in favour of the hypothesis.
When designing artificial intelligent systems, one could do worse, at first glance, than take inspiration from the system whose performance one tries to match: the human brain. The continuing failure to produce such an inspired system is usually blamed on the lack of computational power and/or a lack of understanding of the neuroscience itself. This does not, however, affect the fundamental interest in neuroscience as studying the only known mechanism to date to have produced an intelligent system.
This paper adds another consideration (to the well-established observation that our knowledge of how the brain works is sketchy at best) which needs to be taken into account when taking inspiration from neuroscience: the human brain has evolved specifically to serve the human body under constraints imposed by both the body and biological limitations. This does not necessarily imply that it is futile to consider neuroscience in such endeavours; however, this paper argues that one has to view results of neuroscience from a somewhat different perspective to maximise their utility in the creation of artificial intelligent systems and proposes an explicit separation of neural processes into three categories.
Given that original purpose of cognitive architectures was to lead to a unified theory of cognition, this chapter considers the possible contributions that cognitive architectures can make to embodied theories of cognition in particular. This is not a trivial question since the field remains very much divided about what embodied cognition actually means, and we will see some example positions in this chapter. It is then argued that a useful embodied cognitive architecture would be one that can demonstrate (a) what precisely the role of the body in cognition actually is, and (b) whether a body is constitutively needed at all for some (or all) cognitive processes. It is proposed that such questions can be investigated if the cognitive architecture is designed so that consequences of varying the precise embodiment on higher cognitive mechanisms can be explored. This is in contrast with, for example, those cognitive architectures in robotics that are designed for specific bodies first; or architectures in cognitive science that implement embodiment as an add-on to an existing framework (because then, that framework is by definition not constitutively shaped by the embodiment). The chapter concludes that the so-called semantic pointer architecture by Eliasmith and colleagues may be one framework that satisfies our desiderata and may be well-suited for studying theories of embodied cognition further.
We present initial work on a biologically and cognitively inspired model that may allow embodied agents to autonomously learn sequences of action primitives (forming an overall behaviour). Specifically, we combine a flexible model of sequence generation with a model of parietal mirror neuron activity. The main purpose is to illustrate that the approach is viable. Although further work is needed to improve the results sketched out here, the concept is sound and relevant both to efforts in modelling mirror neuron activity and enabling artificial embodied agents to autonomously learn sequences of action primitives.
Neuroscientific and psychological data suggest a close link between affordance and mirror systems in the brain. However, we still lack a full understanding of both the individual systems and their interactions. Here, we propose that the architecture and functioning of the two systems is best understood in terms of two challenges faced by complex organisms, namely: (a) the need to select among multiple affordances and possible actions dependent on context and high-level goals and (b) the exploitation of the advantages deriving from a hierarchical organisation of behaviour based on actions and action-goals. We first review and analyse the psychological and neuroscientific literature on the mechanisms and processes organisms use to deal with these challenges. We then analyse existing computational models thereof. Finally we present the design of a computational framework that integrates the reviewed knowledge. The framework can be used both as a theoretical guidance to interpret empirical data and design new experiments, and to design computational models addressing specific problems debated in the literature. © 2013 Elsevier Ltd.
The work presented in this paper builds on previous research which analysed human action segmentation in the case of simple object manipulations with the hand (rather than larger-scale actions). When designing algorithms to segment observed actions, for instance to train robots by imitation, the typical approach involves non-linear models but it is less clear whether human action segmentation is also based on such analyses. In the present paper, we therefore explore (1) whether linear models built from observed kinematic variables of a human hand can accurately predict human action segmentation and (2) what kinematic variables are the most important in such a task. In previous work, we recorded speed, acceleration and change in direction for the wrist and the tip of each of the five fingers during the execution of actions as well as the segmentation of these actions into individual components by humans. Here, we use this data to train a large number of models based on every possible training set available and find that, amongst others, the speed of the wrist as well as the change in direction of the index finger were preferred in models with good performance. Overall, the best models achieved R2 values over 0.5 on novel test data but the average performance of trained models was modest. We suggest that this is due to a suboptimal training set (which was not specifically designed for the present task) and that further work be carried out to identify better training sets as our initial results indicate that linear models may indeed be a viable approach to predicting human action segmentation.
In the context of interactive and automated vehicles, driver situation awareness becomes an increasingly important consideration for future traffic systems, whether it concerns the current status of the vehicle or the surrounding environment. Here, we present a simulator study investigating whether the apparent intelligence - i.e. intelligence as perceived by the driver, which is distinct from how intelligent a designer might think the system is - of a vehicle is a factor in the expectations and behaviour of the driver. We are specifically interested in perceived intelligence as a factor in situation awareness. To this end, the study modulates both traffic conditions and the type of navigational assistance given in a goal-navigation task to influence participant's perception of the system. Our result show two distinct effects relevant to situation awareness: 1) Participants who think the vehicle is highly intelligent spend more time glancing at the surrounding environment through the left door window than those who rank intelligence low and 2) participants prefer an awareness of why the navigation aid decided for specific directions but are sensitive to the manner it is presented. Our results have broader implications for the design of future automated systems in vehicles.
Interference between one cognitive behavior or sensory stimulus and subsequent behaviors is a commonly observed effect in the study of human cognition and Psychology. Traditional connectionist approaches explain this phenomenon by mutually inhibiting neural populations underlying those behaviors. Here, we present an alternative model, relying on a more detailed use of synaptic dynamics, in which populations of purely excitatory neurons can nonetheless interfere with each other, causing inhibition of activation for a varying amount of time. The fundamental, biologically motivated, mechanism in the model relies on current “spilling over” from an active neural population into another one, thereby depleting the latter population’s synaptic resources. The principles underlying the model may find applications even in the design of problemsolving artificial neural networks.
We argue that emotions play a central role in human cognition. It is therefore of interest to researchers with an aim to create artificial systems with human-level intelligence (or indeed beyond) to consider the functions of emotions in the human cognition whose complexity they aim to recreate. To this end, we review here several functional roles of emotions in human cognition at different levels, for instance in behavioural regulation and reinforcement learning. We discuss some of the neuroscientific and bodily underpinnings of emotions and conclude with a discussion of possible approaches, including existing efforts, to endow artificial systems with mechanisms providing some of the functions of human emotions. © 2012 Springer-Verlag.
We study how individual components of a complex behavior, so‐called behavioral units, should be sequentially arranged when the overall goal is energy efficiency. We apply an optimization scheme to an existing probabilistic model of C. elegans chemical gradient navigation and find a family of solutions that share common properties. This family is used to analyze general principles of behavioral unit organization, which give rise to search strategies that match qualitatively with those observed in the animal. Specifically, the reorientation behavior emerging in energy efficient virtual worm searchers mimics the pirouette strategy observed in C. elegans, and the virtual worms dwell at the peak of the gradient. Our model predicts that pirouettes are in part associated with the inability to evaluate the gradient during a turn and that the animal does not act upon gradient information while reversing. Together, our results indicate that energy efficiency is an important factor in determining C. elegans gradient navigation. Our framework for the analysis of complex behaviors may, in the future, be used as part of an integrated approach to studying the neural basis of these behaviors.
Robot-assisted therapy (RAT) is an emerging field that has already seen some success and is likely to develop in the future. One particular application area is within therapies for autism spectrum disorders, in which the viability of the approach has been demonstrated.
The present paper is a vision paper with the aim of identifying research directions in the near future of RAT. Specifically, we argue that the next step in such therapeutic scenarios is the development of more substantial levels of autonomy which would allow the robot to adapt to the individual needs of children over longer periods of time (while remaining under the ultimate supervision of a therapist). We argue that this requires new advances on the level of robot controllers as well as the ability to infer and classify intentions, goals and emotional states of the robot’s interactants. We show that the state of the art in a number of relevant disciplines is now at the point at which such an endeavour can be approached in earnest.
In this contribution, we briefly examine the role of end users in the evaluation and characterisation of sophisticated AI-based systems, such as autonomous vehicles or near-future robots. Indeed, when trying to ensure the safety of learning, perception and control in real world settings, one aspect that needs consideration is that human end users are often part of such settings.
We argue that current approaches for considering end users in this respect are insufficient, not the least from a safety perspective, and that this insufficiency will become more acute when transitioning to neuromorphic and/or strongly cognitively inspired solutions. We demonstrate this by borrowing examples from the field of enactivism, which demonstrate that human end users might change the system dynamics of advanced neuromorphic systems when interacting with them, which needs to be taken into consideration. Enactivism might also provide clues as to how to design future evaluation metrics for human-machine teams.
An increasing number of vehicles provide feedbackon eco-driving; information whose purpose it is to increase fuel efficiency in driving. The development of this feedback is relatively novel and there are currently no standards or long-term insights into best design strategies (e.g. leading to a permanent improvement in driving style).
In this paper, we discuss the unexplored relevance of situation awareness (SA) research for eco-driving feedback, highlighting in particular that eco-driving feedback has to be understood as intricately tied into SA. Specifically, we argue that, for the purpose of promoting eco-friendly driving behaviour, the relevant information needs to be part of a driver SA without interfering with safety-critical aspects. We show that this requires eco-feedback systems to possess, themselves, a SA of the traffic situation. This lends support to arguments that SA in a road traffic context is different from SA in the military or aviation domain and takes on a more distributed nature. We conclude by suggesting that head-up displays are a particularly promising interface technology with which to implement the suggestions provided here.
This paper presents a large-scale simulator study on driver adherence to recommendationsgiven by driver support systems, specifically eco-driving support and navigation support.123 participants took part in this study, and drove a vehicle simulator through a pre-defined environment for a duration of approximately 10 min. Depending on the experi-mental condition, participants were either given no eco-driving recommendations, or asystem whose provided support was either basic (recommendations were given in theform of an icon displayed in a manner that simulates a heads-up display) or informative(the system additionally displayed a line of text justifying its recommendations). A naviga-tion system that likewise provided either basic or informative support, depending on thecondition, was also provided.
Effects are measured in terms of estimated simulated fuel savings as well as engine brak-ing/coasting behaviour and gear change efficiency. Results indicate improvements in allvariables. In particular, participants who had the support of an eco-driving system spenta significantly higher proportion of the time coasting. Participants also changed gears atlower engine RPM when using an eco-driving support system, and significantly more sowhen the system provided justifications. Overall, the results support the notion that pro-viding reasons why a support system puts forward a certain recommendation improvesadherence to it over mere presentation of the recommendation.
Finally, results indicate that participants’ driving style was less eco-friendly if the navi-gation system provided justifications but the eco-system did not. This may be due to par-ticipants considering the two systems as one whole rather than separate entities withindividual merits. This has implications for how to design and evaluate a given driver sup-port system since its effectiveness may depend on the performance of other systems in thevehicle.