This work presents an empirical study of repeated robot deception and its effects on changes in behavior and trust in a human-robot interaction scenario. 715 online and 50 in-person participants completed a multitrial driving simulation in which the car's robot assistant repeatedly lies and apologizes. Through a mixed-method approach, our results show that apologies that offer justifications for deception in our scenario mitigate the negative effects on trust over multiple trials. However, given the time-sensitive, high-risk nature of our scenario, none of the apologies caused people to significantly change their decision to exceed the speed limit while rushing their dying friend to the hospital. These results add much needed knowledge to the understudied area of robot deception and could inform designers and policymakers of future practices when considering deploying robots that may learn to deceive.

For social robots to successfully integrate into daily life in home environments, they will need reliable models of the way people perceive and use space in the home. This paper explores the problem of obtaining annotated training data at scale for subjective judgments about spatial locations. Focusing on the use case of identifying good and bad parking spots for a social robot operating in a home environment, two experiments are presented. The first study shows that the presentation of context-rich 3D images to human annotators yields notably different outcomes from those obtained when using 2D robot navigation maps. We attribute the source of these differences to a set of features visible only in the 3D views and introduce a technique for labeling these features on the 2D maps. The second study reveals that using labeled 2D maps produces annotation data very similar to that obtained using 3D images. Since a labeled 2D map can be generated at a fraction of the cost of a full set of 3D views, we recommend this method as a scalable approach to collecting subjective spatial data annotations in everyday environments.

This was the first of a series of experiments to investigate whether adults respond to a robot’s feedback as effectively as to a human’s during a virtual-reality game. Twenty adults volunteered to pop bubbles in 7 virtual-reality games, including 2 in baseline, 3 in acquisition, and 2 in extinction. Participants were randomized to receive feedback from a robot or a human during acquisition. The 80% of the average movement time at baseline was used as the target time to determine the feedback content. Feedback was the same except for the agent (robot or human). Results indicated adults responding to a robot’s feedback to shape their reaching during acquisition and extinction had faster reaching than baseline; however, their patterns were jerkier, less straight, and slower than participants’ receiving a human’s feedback. This study shows the potential to use robots as the feedback provider, though participants performed slightly better with human feedback.

If a robot tells you it can lie for your benefit, how would that change how you perceive it? This paper presents a mixed-methods empirical study that investigates how disclosure of deceptive or honest capabilities influences the perceived social intelligence and construal level of a robot. We first conduct a study with 198 Mechanical Turk participants, and then a replication of it with 15 undergraduate students in order to gain qualitative data. Our results show that how a robot introduces itself can have noticeable effects on how it is perceived–even from just one exposure. In particular, when revealing having ability to lie when it believes it is in the best interest of a human, people noticeably find the robot to be less trustworthy than a robot that conceals any honesty aspects or reveals total truthfulness. Moreover, robots that are forthcoming with their truthful abilities are seen in a lower construal than one that is transparent about its deceptive abilities. These results add much needed knowledge to the understudied area of robot deception and could inform designers and policy makers of future practices when considering deploying robots that deceive.

Previous research has shown that robot mistakes or malfunctions have a significant negative impact on people’s trust. One way to mitigate the negative impact of trust violation is through trust repair. Although trust repair has been studied extensively, it is still not known which strategy is effective in repairing trust in a time-sensitive driving scenario. Additionally, prior research on trust repair has not dealt with the effects of expressing emotion in attempting trust repair. In this paper, we presented the development of a variety of trust repair methods for a time-sensitive scenario using a simulated driving environment as a testbed for validation. These trust repair methods included baseline apology, emotional apology, and explanation. We conducted an experiment to compare the impact of these trust repair methods on human-robot trust. Experimental results indicated that the emotional apology positively affected more participants than the no-repair, baseline apology, and explanation. Furthermore, this study identified emotional apology as the most effective method for the time-sensitive driving scenario.

The growing potential for facial emotion recognition (FER) technology has encouraged expedited development at the cost of rigorous validation. Many of its use-cases may also impact the diverse global community as FER becomes embedded into domains ranging from education to security to healthcare. Yet, prior work has highlighted that FER can exhibit both gender and racial biases like other facial analysis techniques. As a result, bias-mitigation research efforts have mainly focused on tackling gender and racial disparities, while other demographic related biases, such as age, have seen less progress. This work seeks to examine the performance of state of the art commercial FER technology on expressive images of men and women from three distinct age groups. We utilize four different commercial FER systems in a black box methodology to evaluate how six emotions - anger, disgust, fear, happiness, neutrality, and sadness - are correctly detected by age group. We further investigate how algorithmic changes over the last year have affected system performance. Our results found that all four commercial FER systems most accurately perceived emotion in images of young adults and least accurately in images of older adults. This trend was observed for analyses conducted in 2019 and 2020. However, little to no gender disparities were observed in either year. While older adults may not have been the initial target consumer of FER technology, statistics show the demographic is quickly growing more keen to applications that use such systems. Our results demonstrate the importance of considering various demographic subgroups during FER system validation and the need for inclusive, intersectional algorithmic developmental practices.

Wearable IMU sensing systems have widely been used in the study of human motion. For example, gait analysis using wearable inertial sensor systems is a tool used by clinicians to discriminate between typical and pathological walking. Similarly, key descriptors can be identified in spontaneous kicking to distinguish between typical and atypical motor development in infants. Oftentimes in human applications, precise placement of inertial sensors is difficult due to the irregular shape of human limbs. Without precise placement and alignment of the inertial sensors, meaningful joint kinematic data are difficult to extract as the orientation of the joint axes are unknown in the sensor's local frame. So, for applications where precise alignment may not be possible, a necessary first step is to identify the joint axes with respect to the local frame.In this work, we propose a method for the identification of joint axes for multiple degree of freedom (multi-DOF) joints in a kinematic chain using acceleration and angular rate data. This method couples a thresholding activity detection algorithm with a principal component analysis (PCA) dimensionality reduction technique. Furthermore, this method is validated on mimicked kicking data from a NAO robot. This method can determine joint axes of a kinematic chain from simultaneous movement data within an error ratio of 0.09.

Using robots in therapy for children on the autism spectrum is a promising avenue for child-robot interaction, and one that has garnered significant interest from the research community. After preliminary interviews with stakeholders and evaluating music selections, twelve typically developing (TD) children and three children with Autism Spectrum Disorder (ASD) participated in an experiment where they played the dance freeze game to four songs in partnership with either a NAO robot or a human partner. Overall, there were significant differences between TD children and children with ASD (e.g., mimicry, dance quality, & game play). There were mixed results for TD children, but they tended to show greater engagement with the researcher. However, objective results for children with ASD showed greater attention and engagement while dancing with the robot. There was little difference in game performance between partners or songs for either group. However, upbeat music did encourage greater movement than calm music. Using a robot in a musical dance game for children with ASD appears to show the advantages and potential just as in previous research efforts. Implications and future research are discussed with the results.

The attribution of human-like characteristics onto humanoid robots has become a common practice in Human-Robot Interaction by designers and users alike. Robot gendering, the attribution of gender onto a robotic platform via voice, name, physique, or other features is a prevalent technique used to increase aspects of user acceptance of robots. One important factor relating to acceptance is user trust. As robots continue to integrate themselves into common societal roles, it will be critical to evaluate user trust in the robot's ability to perform its job. This paper examines the relationship among occupational gender-roles, user trust and gendered design features of humanoid robots. Results from the study indicate that there was no significant difference in the perception of trust in the robot's competency when considering the gender of the robot. This expands the findings found in prior efforts that suggest performance-based factors have larger influences on user trust than the robot's gender characteristics. In fact, our study suggests that perceived occupational competency is a better predictor for human trust than robot gender or participant gender. As such, gendering in robot design should be considered critically in the context of the application by designers. Such precautions would reduce the potential for robotic technologies to perpetuate societal gender stereotypes.

Intelligent systems, especially those with an embodied construct, are becoming pervasive in our society. From chatbots to rehabilitation robotics, from shopping agents to robot tutors, people are adopting these systems into their daily life activities. Alas, associated with this increased acceptance is a concern with the ethical ramifications as we start becoming more dependent on these devices [1]. Studies, including our own, suggest that people tend to trust, in some cases overtrusting, the decision-making capabilities of these systems [2]. For high-risk activities, such as in healthcare, when human judgment should still have priority at times, this propensity to overtrust becomes troubling [3]. Methods should thus be designed to examine when overtrust can occur, modelling the behavior for future scenarios and, if possible, introduce system behaviors in order to mitigate. In this talk, we will discuss a number of human-robot interaction studies conducted where we examined this phenomenon of overtrust, including healthcare-related scenarios with vulnerable populations, specifically children with disabilities.

The kicking patterns of infants can provide markers that may predict the trajectory of their future development. Atypical kicking patterns may predict the possibility of developmental disorders like Cerebral Palsy (CP). Early intervention and physical therapy that encourages the practice of proper kicking motions can help to improve the outcomes in these scenarios. The kicking motions of an infant are usually evaluated by a trained health professional and subsequent physical therapy is also conducted by a licensed professional. The automation of the evaluation of kicking motions and the administration of physical therapy is desirable for standardizing these processes. In this work, we attempt to develop a method to quantify metrics that can provide insight into the quality of baby kicking actions. We utilize a computer vision system to analyze infant kicking stimulated by parent-infant play and a robotic infant mobile. We utilize statistical techniques to estimate kick type (synchronous and non-synchronous), kick amplitude, kick frequency, and kick deviation. These parameters can prove helpful in determining an infant's kick quality and also measure improvements in physical therapy over time. In this paper, we detail the design of the system and discuss the statistical results.

Trust is a key element for successful human-robot interaction. One challenging problem in this domain is the issue of how to construct a formulation that optimally models this trust phenomenon. This paper presents a framework for modeling human-robot trust based on representing the human decision-making process as a formulation based on trust states. Using this formulation, we then discuss a generalized model of human-robot trust based on Hidden Markov Models and Logistic Regression. The proposed approach is validated on datasets collected from two different human subject studies in which the human is provided the ability to take advice from a robot. Both experimental scenarios were time-sensitive, in that a decision had to be made by the human in a limited time period, but each scenario featured different levels of cognitive load. The experimental results demonstrate that the proposed formulation can be utilized to model trust, in which the system can predict whether the human will decide to take advice (or not) from the robot. It was found that our prediction performance degrades after the robot made a mistake. The validation of this approach on two scenarios implies that this model can be applied to other interactive scenarios as long as the interaction dynamics fits into the proposed formulation. Directions for future improvements are discussed.

Embodiment of actions and tasks has typically been analyzed from the robot's perspective where the robot's embodiment helps develop and maintain trust. However, we ask a similar question looking at the interaction from the human perspective. Embodied cognition has been shown in the cognitive science literature to produce increased social empathy and cooperation. To understand how human embodiment can help develop and increase trust in human-robot interactions, we created conducted a study where participants were tasked with memorizing greek letters associated with dance motions with the help of a humanoid robot. Participants either performed the dance motion or utilized a touch screen during the interaction. The results showed that participants' trust in the robot increased at a higher rate during human embodiment of motions as opposed to utilizing a touch screen device.

Infant kicking patterns can provide clues for causes for concern with future development. Cerebral Palsy is a development disorder that may be predicted by observing the spontaneous kicking patterns of an infant. Early detection and intervention can improve the overall long term outcome through specific physical therapy exercises. Since all infants are unique it may be beneficial to learn child specific stimuli that optimize the quantity of kicking actions. Discovering the stimuli that will encourage a particular infant to perform kicking actions will give healthcare professionals more opportunities to observe and evaluate these actions for possible atypical patterns. We expand on previous work that utilizes computer vision and a robotic baby mobile that detects infant kicking motions and activates stimuli to encourage continued kicking. Based on the observed state-action pairs recorded while an infant interacts with the robotic baby mobile, we develop a Markov Decision Process and calculate an optimal policy to encourage an increased amount of kicking. This method could theoretically be applied to different infants, which would result in varying optimal policies that are specific to each child. In this paper we will briefly describe the robotic system, discuss the resulting Markov Decision Process and optimal policy, and describe future works.

Trust is an important characteristic of successful interactions between humans and agents in many scenarios. Self-driving scenarios are of particular relevance when discussing the issue of trust due to the high-risk nature of erroneous decisions being made. The present study aims to investigate decision-making and aspects of trust in a realistic driving scenario in which an autonomous agent provides guidance to humans. To this end, a simulated driving environment based on a college campus was developed and presented. An online and an in-person experiment were conducted to examine the impacts of mistakes made by the self-driving AI agent on participants' decisions and trust. During the experiments, participants were asked to complete a series of driving tasks and make a sequence of decisions in a time-limited situation. Behavior analysis indicated a similar relative trend in the decisions across these two experiments. Survey results revealed that a mistake made by the self-driving AI agent at the beginning had a significant impact on participants' trust. In addition, similar overall experience and feelings across the two experimental conditions were reported. The findings in this study add to our understanding of trust in human-robot interaction scenarios and provide valuable insights for future research work in the field of human-robot trust.

The diagnosis of Autism Spectrum Disorder (ASD) in children is commonly accompanied by a diagnosis of sensory processing disorders. Abnormalities are usually reported in multiple sensory processing domains, showing a higher prevalence of unusual responses, particularly to tactile, auditory and visual stimuli. This paper discusses a novel robot-based framework designed to target sensory difficulties faced by children with ASD in a controlled setting. The setup consists of a number of sensory stations, together with two different robotic agents that navigate the stations and interact with the stimuli. These stimuli are designed to resemble real world scenarios that form a common part of one's everyday experiences. Given the strong interest of children with ASD in technology in general and robots in particular, we attempt to utilize our robotic platform to demonstrate socially acceptable responses to the stimuli in an interactive, pedagogical setting that encourages the child's social, motor and vocal skills, while providing a diverse sensory experience. A preliminary user study was conducted to evaluate the efficacy of the proposed framework, with a total of 18 participants (5 with ASD and 13 typically developing) between the ages of 4 and 12 years. We derive a measure of social engagement, based on which we evaluate the effectiveness of the robots and sensory stations in order to identify key design features that can improve social engagement in children.

In recent news, organizations have been considering the use of facial and emotion recognition for applications involving youth such as tackling surveillance and security in schools. However, the majority of efforts on facial emotion recognition research have focused on adults. Children, particularly in their early years, have been shown to express emotions quite differently than adults. Thus, before such algorithms are deployed in environments that impact the wellbeing and circumstance of youth, a careful examination should be made on their accuracy with respect to appropriateness for this target demographic. In this work, we utilize several datasets that contain facial expressions of children linked to their emotional state to evaluate eight different commercial emotion classification systems. We compare the ground truth labels provided by the respective datasets to the labels given with the highest confidence by the classification systems and assess the results in terms of matching score (TPR), positive predictive value, and failure to compute rate. Overall results show that the emotion recognition systems displayed subpar performance on the datasets of children's expressions compared to prior work with adult datasets and initial human ratings. We then identify limitations associated with automated recognition of emotions in children and provide suggestions on directions with enhancing recognition accuracy through data diversification, dataset accountability, and algorithmic regulation.

Several HRI studies have investigated the use of interactive social robots to enhance a variety of activities designed for children in recent years. Notably, these robots have shown the ability to facilitate greater engagement with rehabilitative therapy exercises. Yet, very few studies have directly compared the performance of a robotic therapist with that of a human therapist during therapy sessions. This type of analysis is even less common for studies involving children. This work presents the experimental results of a between subjects study conducted with 10 children who interacted with either a robot or human therapist while playing a virtual reality rehabilitation game. The results are analyzed in terms of user intrinsic motivation and compared to previous work with adult participants. Preliminary results show a trend that children enjoyed working with the robot therapist more than the human therapist while playing the game. This finding is consistent with related work within the domain of child-robot interaction and supports the continued investigation of the potential uses of social robots in therapeutic rehabilitation protocols for children.

Dementia is a growing problem amongst elderly adults and the number of dementia patients is predicted to rise considerably in the coming years. While there is no cure for dementia, recent studies have suggested that exercise may have a positive effect on the cognitive function of dementia patients. We propose that a humanoid therapy robot is an effective tool for encouraging exercise in dementia patients. Such a robot will help address problems such as cost of care and shortage of healthcare workers. We have developed an interactive robotic system and conducted preliminary tests with a robot that encourages a user to engage in simple dance moves. The heart rate is used as feedback to decide which exercise move should be demonstrated. The results we have found are promising and we hope to continue this work via future studies.

According to data from the Centers for Disease Control and Prevention, developmental disorders such as Autism Spectrum Disorder (ASD) and Cerebral Palsy (CP) affect nearly one in six children between the ages of 3 to 17 in the United States alone. In order to improve the quality of life for these individuals, there is increased emphasis on providing early intervention at infancy, when key developmental milestones are being achieved. This however requires accurate early detection of motor development delays in at-risk infants. Our research focuses on enabling early detection through the design of a robotic crib mobile that affects infant behavior. Stimuli integrated into the robotic mobile can be used to encourage certain motions such as kicking among infants in order to study infant motor development and identify at-risk populations. In this paper, we propose the design of such a robotic crib mobile and discuss preliminary results from deploying the mobile in the infants' home environment during a pilot study.

Collaborative robots that work alongside humans will experience service breakdowns and make mistakes. These robotic failures can cause a degradation of trust between the robot and the community being served. A loss of trust may impact whether a user continues to rely on the robot for assistance. In order to improve the teaming capabilities between humans and robots, forms of communication that aid in developing and maintaining trust need to be investigated. In our study, we identify four forms of communication which dictate the timing of information given and type of initiation used by a robot. We investigate the effect that these forms of communication have on trust with and without robot mistakes during a cooperative task. Participants played a memory task game with the help of a humanoid robot that was designed to make mistakes after a certain amount of time passed. The results showed that participants' trust in the robot was better preserved when that robot offered advice only upon request as opposed to when the robot took initiative to give advice.

Neuro disorders such as Cerebral Palsy (CP) and Infantile Spasms (IS) in infants can cause a wide range of developmental coordination disorders (DCD). Simultaneous, non-complex, kicking patterns that persist in infants of 4-7 months of age is highly suggestive of a neuro-disorder. Early establishment of risk levels for infants at risk for neuro-disorders is beneficial for early intervention. To provide a method to track early infant kicking movements for determining risk-level, an automated method is established to track and classify periods of simultaneous (SM), non-simultaneous movements (NSM) and no movement (NM) during infant kicking actions. In this paper, a computer vision algorithm uses KAZE points to track infant kicking and collect kinematic data. Each movement type is classified by computing unique feature criterion and using a support vector machine (SVM) for learning a movement model. We discuss the significance of the classifier as well as analyze the percentage break down of movement types for typical infants and infants with IS.

With the recent advances in computing, artificial intelligence (AI) is quickly becoming a key component in the future of advanced applications. In one application in particular, AI has played a major role - that of revolutionizing traditional healthcare assistance. Using embodied interactive agents, or interactive robots, in healthcare scenarios has emerged as an innovative way to interact with patients. As an essential factor for interpersonal interaction, trust plays a crucial role in establishing and maintaining a patient-agent relationship. In this paper, we discuss a study related to healthcare in which we examine aspects of trust between humans and interactive robots during a therapy intervention in which the agent provides corrective feedback. A total of twenty participants were randomly assigned to receive corrective feedback from either a robotic agent or a human agent. Survey results indicate trust in a therapy intervention coupled with a robotic agent is comparable to that of trust in an intervention coupled with a human agent. Results also show a trend that the agent condition has a medium-sized effect on trust. In addition, we found that participants in the robot therapist condition are 3.5 times likely to have trust involved in their decision than the participants in the human therapist condition. These results indicate that the deployment of interactive robot agents in healthcare scenarios has the potential to maintain quality of health for future generations.

The use of socially interactive robots in people's homes and workplaces will continue to grow. To test such systems before deployment, most human-robot interaction experiments are usually conducted in a research laboratory or require the physical presence of the robot. These experiments are usually time-consuming, costly, and, in most cases, require an experimenter to be present. To address this problem, we have developed a web-based research platform in which participants can remotely interact with a socially interactive robot shown via video based on a behavior profile programmed via a Wizard-of-Oz framework. To help us validate the believability when using this platform, we compared participant's responses to a physical presence robot with that to a remote presence robot during a problem-solving scenario. Participants were asked to complete a series of cognitive tasks with the help of the robot. We analyzed participant's tendency to take the robot's advice, their feelings before and after the experiment, and a quantitative assessment about believability. The experimental results indicate that there was a high degree of agreement between participants' behaviors in the two conditions and only minor changes in feelings before and after the experiment. The survey responses found no significant difference in their overall rating for believability, suggesting the equivalency of believability between the two conditions.

With recent advances in robotics, it is expected that robots will become increasingly common in human environments, such as in the home and workplaces. Robots will assist and collaborate with humans on a variety of tasks. During these collaborations, it is inevitable that disagreements in decisions would occur between humans and robots. Among factors that lead to which decision a human should ultimately follow, theirs or the robot, trust is a critical factor to consider. This study aims to investigate individuals' behaviors and aspects of trust in a problem-solving situation in which a decision must be made in a bounded amount of time. A between-subject experiment was conducted with 100 participants. With the assistance of a humanoid robot, participants were requested to tackle a cognitive-based task within a given time frame. Each participant was randomly assigned to one of the following initial conditions: 1) a working robot in which the robot provided a correct answer or 2) a faulty robot in which the robot provided an incorrect answer. Impacts of the faulty robot behavior on participant's decision to follow the robot's suggested answer were analyzed. Survey responses about trust were collected after interacting with the robot. Results indicated that the first impression has a significant impact on participant's behavior of trusting a robot's advice during a disagreement. In addition, this study discovered evidence supporting that individuals still have trust in a malfunctioning robot even after they have observed a robot's faulty behavior.

Many of us, roboticists and those who collaborate with them, experience delight, excitement, and sometimes deep-seated, but rarely unvoiced, fears as we witness our robotic systems begin to impact human lives in countless ways.From automating driving to reshaping various facets of health care delivery, robotic systems are growing in their prevalence and intrusiveness into our daily lives.In combination with our siblings in the Artificial Intelligence (AI) community, scholars continue to predict a wide range of benefits from robotics and AI systems but also serious harms, including potential existential threats to humanity.Recognized pillars of science and engineering, including Elon Musk and the late great Stephen Hawking, have given voice to the apocalyptic kinds of fears that the public may have about an increasingly automated future.Whether these fears should be taken seriously is an issue that has divided scholars for awhile now, as illustrated by debates between Bill Joy [7] and Ray Kurzweil [8] at the beginning of the 21st century.On a different scale of granularity, a category of harms that users and others are more likely to experience on a day-to-day basis is from the various types of bias encoded in, or learned by, AI systems.This category of harms is especially troublesome in the world of physical robotics.Nonembodied AI systems can obviously make decisions that have effects on human beings, such as a chatbot determining what to say in response to a customer's question on a company's helpline.Yet it will need to rely on an embodied entity (often a human) to have a direct impact on the physical world.Typically, a nonembodied AI agent provides input to humans who may then execute a physical action -whether those humans are making an employment decision to hire or fire or deciding on a health care intervention for a patient.By definition, it lacks the capability of acting on the world without assistance.Robots that have a physical form, on the other hand, can perform actions on their own.This can raise the ethical stakes in terms of the potential benefits and harms that may result from the technology.The benefits and harms that we are particularly concerned about here are related to bias.

This article presents a conceptual framework for human-robot trust which uses computational representations inspired by game theory to represent a definition of trust, derived from social psychology. This conceptual framework generates several testable hypotheses related to human-robot trust. This article examines these hypotheses and a series of experiments we have conducted which both provide support for and also conflict with our framework for trust. We also discuss the methodological challenges associated with investigating trust. The article concludes with a description of the important areas for future research on the topic of human-robot trust.

As the technology needed to make robots robust and affordable draws ever nearer, human-robot interaction (HRI) research to make robots more useful and accessible to the general population becomes more crucial. In this study, 59 college students filled out an online survey soliciting their judgments regarding seven social robots based solely on appearance. Results suggest that participants prefer robots that resemble animals or humans over those that are intended to represent an imaginary creature or do not resemble a creature at all. Results are discussed based on social robot application and design features.

Physical embodiment of robots provides users with a social environment. To design social robots further to be accepted as our companions, we need to understand the essential factors and implement them and so, users get to bring them to their personal environments. To this aim, we focused on two important factors in robotic companionship: robot appearance (look) and emotional expression (feel). Twenty-one participants played an online game with the help from two humanoid robots, Nao (more human-like looking) and Darwin (less human-like looking). Participants interacted with each robot either with emotional words or without emotional words. Results show that only when the robot both looks more human-like and speaks with emotional expression, participants perceive it as their companion. Implications are discussed with future works.

Robots have the potential to save lives in high-risk situations, such as emergency evacuations. To realize this potential, we must understand how factors such as the robot's performance, the riskiness of the situation, and the evacuee's motivation influence his or her decision to follow a robot. In this paper, we developed a set of experiments that tasked individuals with navigating a virtual maze using different methods to simulate an evacuation. Participants chose whether or not to use the robot for guidance in each of two separate navigation rounds. The robot performed poorly in two of the three conditions. The participant's decision to use the robot and self-reported trust in the robot served as dependent measures. A 53% drop in self-reported trust was found when the robot performs poorly. Self-reports of trust were strongly correlated with the decision to use the robot for guidance (φ(90) = +0.745). We conclude that a mistake made by a robot will cause a person to have a significantly lower level of trust in it in later interactions.

Children with autism have a hard time both with expressing their emotions, and with understanding the emotions of others. As the population of children with autism increases, it is crucial we create effective therapeutic programs that will improve their communication skills. We present an interactive robotic system that delivers emotional and social behaviors for multi-sensory therapy for children with autism spectrum disorders. Our framework includes emotion-based robotic gestures and facial expressions, as well as vision and audio-based monitoring system for quantitative measurement of the interaction.

We present an interactive robotic framework that delivers emotional and social behaviors for multi-sensory therapy for children with autism spectrum disorders. Our framework includes emotion-based robotic gestures and facial expressions, as well as vision and audio-based monitoring system for quantitative measurement of the interaction. We also discuss the special aspects of interacting with children with autism with multi-sensory stimuli and the potentials of our approach for personalized therapies for social and behavioral learning.

Robots have the potential to save lives in emergency scenarios, but could have an equally disastrous effect if participants overtrust them. To explore this concept, we performed an experiment where a participant interacts with a robot in a non-emergency task to experience its behavior and then chooses whether to follow the robot's instructions in an emergency or not. Artificial smoke and fire alarms were used to add a sense of urgency. To our surprise, all 26 participants followed the robot in the emergency, despite half observing the same robot perform poorly in a navigation guidance task just minutes before. We performed additional exploratory studies investigating different failure modes. Even when the robot pointed to a dark room with no discernible exit the majority of people did not choose to safely exit the way they entered.

Previous studies have shown that external corrective feedback provided to individuals undergoing physical therapy sessions increases the efficacy of their intervention protocols, thus allowing for the individual to rapidly improve on their performance. However, direct feedback is typically provided by an expert therapist during weekly or monthly visits, which limits improvement on a daily basis. As such, we promote in-home rehabilitation protocols by developing a novel framework that couples serious games with a robotic playmate. After showing in a previous study that a combination of verbal and nonverbal cues is the most efficient method for providing guided instruction, we now show that the efficacy of the intervention protocols can be further increased by embedding human-like behaviors on the robotic platform such that it provides continuous feedback. This would allow user performance to improve at each iteration of feedback and, ultimately, reach an individualized performance goal. To show that users can reach a reference point for a given kinematic parameter, we recruited adults and children to interact with our system. Results from the first round of experiments show that adults who received feedback from an embodied agent need less trials on average to reach their performance goals than the adults who received feedback from a virtual agent. Results from the second round of experiments further support our hypothesis by showing that children also reach their performance goals by receiving feedback from the robotic playmate. We conclude that coupling existing serious games with an embodied robotic agent has the potential of increasing the efficacy of intervention protocols by allowing for use outside the clinical setting thus increasing users' rate of improvement.

Most Human-Robot Interaction (HRI) experiments are costly and time consuming because they involve deploying a physical robot in a physical space. Experiments using virtual environments can be easier and less expensive, but it is difficult to ensure that the results will be valid in the physical domain. To begin to answer this concern, we have performed an evaluation comparing participants' understanding of robotic guidance instructions using robots that were virtually, remotely, or physically present for the experiment. All but one set of experimental conditions gave similar results across the three presence levels. Further, we find that qualitative responses about the robots were largely the same regardless of presence level.

We explored symbolic canine-human communication for working dogs through the use of canine head gestures. We identified a set of seven criteria for selecting head gestures and identified the first four deserving further experimentation. We devised computationally inexpensive mechanisms to prototype the live system from a motion sensor on the dog's collar. Each detected gesture is paired with a predetermined message that is voiced to the humans by a smart phone. We examined the system and proposed gestures in two experiments, one indoors and one outdoors. Experiment A examined both gesture detection accuracy and a dog's ability to perform the gestures using a predetermined routine of cues. Experiment B examined the accuracy of this system on two outdoor working-dog scenarios. The detection mechanism we presented is sufficient to point to improvements into system design and provide valuable insights into which gestures fulfill the seven minimum criteria.

A robot companion should adapt to its user's needs by learning to perform new tasks. In this paper, we present a robot playmate that learns and adapts to tasks chosen by the child on a touchscreen tablet. We aim to solve the task learning problem using an experience-based learning framework that stores human demonstrations as task instances. These instances are retrieved when confronted with a similar task in which the system generates predictions of task behaviors based on prior actions. In order to automate the processes of instance encoding, acquisition, and retrieval, we have developed a framework that gathers task knowledge through interaction with human teachers. This approach, further referred to as interactive instance-based learning (IIBL), utilizes limited information available to the robot to generate similarity metrics for retrieving instances. In this paper, we focus on introducing and evaluating a new hybrid IIBL framework using sensitivity analysis with artificial neural networks and discuss its advantage over methods using k-NNs and linear regression in retrieving instances.

For individuals with a motor skill disorder, repetition of recommended therapy exercises is essential for motor improvement. Moreover, external feedback of performance is an important component of therapy such that individuals can correct their exercises and improve their performance. However, direct feedback is typically only provided by an expert therapist during weekly or monthly therapy sessions, which limits improvement on a daily basis. In order to promote the repetition of recommended exercises in a home setting, several serious games have been developed to promote compliance with therapy interventions. To advance this work, we have developed a novel framework to couple serious games with a robot playmate that provides corrective feedback during interaction. The playmate continuously tracks the user's kinematic performance and autonomously provides objective verbal and nonverbal cues in order to increase the efficacy of the intervention. To determine how various cues affect an individual's kinematic performance, we have tested the complete system with 20 able-bodied adults. Namely, we computed the total amount of time it took the participants to successfully complete a reaching task as a function of the verbal or nonverbal cues received. The results show that movement times improve at a faster rate for the group provided with both verbal and nonverbal feedback versus verbal feedback alone. Exit surveys also suggest that the system was deemed enjoyable by the targeted population.

The goal of this research is to evaluate the benefits of using a low-cost Virtual Reality (VR) surgical training platform in bridging the gap between early surgical skills and effective performance in more advanced training and clinical settings. The specific aim of this study is to design and evaluate the efficacy of a low-cost virtual reality system as a precursor to improving surgical skills for novice trainees. A summary of our VR training system is presented, and training results from three pre-med students and five residents in a medical school are discussed that show preliminary evidence of the efficacy of the system.

One of the key elements for building a long-term robotic companion is incorporating the ability for a robot to continuously learn and engage in new tasks. Utilizing a defined workspace that provides various shared content between human and robot could assist in this learning process. Here, we propose integrating a touchscreen tablet and a robot learner for engaging the user during human-robot interaction scenarios. The robot learner's domain-independent core reasoner follows the structure of instance-based learning which addresses the issues of acquiring knowledge, encoding cases, and learning a retrieval metric. The system utilizes demonstrations provided by the user to auto-populate the knowledge base through natural interaction methods, encodes cases based on the feature structure provided by the user, and uses an adaptive-weighting technique to design a retrieval metric with linear regression in the feature-distance space. Through a tablet environment, the user teaches a task to the robot in a shared workspace and intuitively monitors the robot's behavior and progress in real time. In this setting, the user is able to interrupt the robot and provide necessary demonstrations at the moment learning is taking place, thus providing a means to continuously engage both the participant and the robot in the learning cycle.

The role of emotions in social scenarios is to provide an inherent mode of communication between two parties. When emotions are properly employed and understood, people are able to respond appropriately, which further enhances the social interaction. Ultimately, effective emotion execution in social settings has the capability to build rapport, improve engagement, optimize learning, provide comfort, and increase overall likability. In this paper, we discuss associating dominant emotions of effective social interaction to gestural behaviors on a humanoid robotic platform. Studies with 13 participants interacting with the robot show that by integrating key principles related to the characteristics of happy and sad emotions, the intended emotion is perceived across all participants with 95.19% and 94.23% sensitivity, respectively.

In the rehabilitation field, determining the effectiveness of an intervention protocol begins by comparing the individual's movement characteristics against a baseline. In most settings, this baseline is determined through clinical studies involving a range of patients belonging to the same demographic group. Unfortunately, this leads to a process that is difficult to repeat for all patient demographics, or all movement characteristics, given the demands on clinicians' and patients' time for performing such clinical baseline measurement studies. To address this issue, we discuss a method that allows clinicians to objectively assess an individual's movements and compare the resulting outcome kinematic metrics to a kinematic baseline. Instead of collecting human patient data, we propose a robotic kinematic model that generates a baseline for different kinematic parameters in real-time as a function of the state of a given task. We evaluate our methodology on elbow and shoulder range of motion (ROM) angles obtained from eleven typically developing children. We compare the user's ROM angles to those generated by the proposed model, and discuss the potential of the model to be used in various intervention protocols.

Motivated by the desire to mitigate human casualties in emergency situations, this paper explores various guidance modalities provided by a robotic platform for instructing humans to safely evacuate during an emergency. We focus on physical modifications of the robot, which enables visual guidance instructions, since auditory guidance instructions pose potential problems in a noisy emergency environment. Robotic platforms can convey visual guidance instructions through motion, static signs, dynamic signs, and gestures using single or multiple arms. In this paper, we discuss the different guidance modalities instantiated by different physical platform constructs and assess the abilities of the platforms to convey information related to evacuation. Human-robot interaction studies with 192 participants show that participants were able to understand the information conveyed by the various robotic constructs in 75.8% of cases when using dynamic signs with multi-arm gestures, as opposed to 18.0% when using static signs for visual guidance. Of interest to note is that dynamic signs had equivalent performance to single-arm gestures overall but drastically different performances at the two distance levels tested. Based on these studies, we conclude that dynamic signs are important for information conveyance when the robot is in close proximity to the human but multi-arm gestures are necessary when information must be conveyed across a greater distance.

Icebergs generated from ice ablation processes continue to be a threat for operations conducted in polar regions. Systems that have been developed to track and observe these threats often use either space-based radar imaging or visual observation by the crew of the ship. Both of these methods have disadvantages, mostly in terms of real-time observation or the physical abilities of the crew. We propose a robotic solution for in-situ observation of icebergs, so that countermeasures may be quickly implemented. Our focus in this work is the problem of allocating resources to observation regions: once areas of iceberg activity have been identified, how are robot observers assigned to these regions and what cost metric may be used to determine the best placement of robot observers. Our solution is currently demonstrated and evaluated in simulation.

Despite advances in underwater technology, human divers are still employed to carry out a wide variety of dangerous underwater tasks. To enable a diver to complete tasks safer and more efficiently, it is proposed that an Underwater Robotic Assistant (UWRA) could act as a robotic dive buddy. For example, the UWRA could ferry tools from the surface, carry scientific samples, and provide illumination in dark environments. However, one of the major hurdles to operating autonomous systems below the water's surface is that high-frequency radio signals are greatly attenuated by water. This severely limits the type of commands a diver can issue to the robot. Thus, one of the contributions of this paper is an analysis of the types of Underwater Human-Robot Communication (UHRC) that are possible between a diver and a UWRA. Following the analysis, we discuss implementation of a few of the proposed methods of UHRC where an underwater robot assists a diver during the installation of an underwater mechanical platform.

Underwater environments hidden beneath a layer of ice prove challenging to explore for both humans and unmanned underwater vehicles. Exploration of such areas is a topic of interest to many scientific communities. In this paper, we discuss details on the approach taken to obtain sonar and video data of under ice topography in a Colorado mountain lake using a VideoRay Pro IV unmanned underwater vehicle. Video from the vehicle along with acoustic sonar data from a BlueView forward looking sonar sensor were recorded concurrently for use in future research. The approach taken, challenges encountered, results obtained, and lessons learned during the data collection process are described herein.

Navigation through underwater environments is challenging given the lack of accurate positioning systems. The determination of underwater vehicle movement using an integrated acoustic sonar sensor would provide underwater vehicles with greatly increased autonomous navigation capabilities. A forward looking sonar sensor may be used for determining autonomous vehicle movement using filtering and optical flow algorithms. Optical flow algorithms have shown excellent results for vision image processing. However, they have been found difficult to implement using sonar data due to the high level of noise present, as well as the widely varying appearances of objects from frame to frame. For the bottom tracking applications considered, the simplifying assumption can be made that all features move with an equivalent direction and magnitude between frames. Statistical analysis of all estimated feature movements provides an accurate estimate of the overall shift, which translates directly to the vehicle movement. Results using acoustic sonar data are presented which illustrate the effectiveness of this methodology.

Virtual reality (VR) surgical training can be a potentially useful method for improving practicing surgical skills. However, the current literature on VR training has not discussed the efficacy of VR systems that are useful outside of the training facility. As such, the goal of this study is to evaluate the benefits of using a low-cost VR simulation system for providing a method to increase the learning of surgical skills. Our pilot case focuses on laparoscopic cholecystectomy, which is one of the most common surgeries currently performed in the United States and is often used as the training case for laparoscopy due to its high frequency and perceived low risk. The specific aim of this study is to examine the efficacy of a low-cost haptic-based VR surgical simulator on improving practicing surgical skills, measured by the change in the learning effect of students.

This paper presents a robotic system that provides telepresence to the visually impaired by combining real-time haptic rendering with multi-modal interaction. A virtual-proxy based haptic rendering process using a RGB-D sensor is developed and integrated into a unified framework for control and feedback for the telepresence robot. We discuss the challenging problem of presenting environmental perception to a user with visual impairments and our solution for multi-modal interaction. We also explain the experimental design and protocols, and results with human subjects with and without visual impairments. Discussion on the performance of our system and our future goals are presented toward the end.

Environmental monitoring of spatially-distributed geo-physical processes (e.g., temperature, pressure, or humidity) requires efficient sampling schemes, particularly, when employing an autonomous mobile agent to execute the sampling task. Many approaches have considered optimal sampling strategies which specialize in minimizing estimation error, while others emphasize reducing resource usage, yet rarely are both of these performance parameters used concurrently to influence the navigation. This work discusses how a spatial estimation process and resource awareness are integrated to generate an informed navigation policy for collecting useful measurement information. We also enable a direct comparison between this informed navigation method and more common approaches using two performance metrics. We show that our informed navigation outperforms these approaches based on performance evaluation as a function of estimation error and resource usage for a useful range of coverage within the sampling area.

Adaptive learning is an educational method that utilizes computers as an interactive teaching device. Intelligent tutoring systems, or educational agents, use adaptive learning techniques to adapt to each student's needs and learning styles in order to individualize learning. Effective educational agents should accomplish two essential goals during the learning process - 1) monitor engagement of the student during the interaction and 2) apply behavioral strategies to maintain the student's attention when engagement decreases. This paper focuses on the second objective of reengaging students using various behavioral strategies through the utilization of a robotic educational agent. Details are provided on the overall system approach and the forms of verbal and nonverbal cues used by the robotic agent. Results derived from 24 students engaging with the robot during a computer-based math test show that, while various forms of behavioral strategies increase test performance, combinations of verbal cues result in a slightly better outcome.

Ships that operate in polar regions continue to face the threat of floating ice sheets and icebergs generated from an ice ablation process. Systems have been implemented to track these threats, with varying degrees of success. In this paper, a definition is proposed for this tracking problem that re-casts it within a class of robotic, multiagent target observation problems. The focus in this new definition is on minimizing the time an initial contact for newly generated targets is obtained, as opposed to obtaining a contact long after a target has been generated. Focusing on the initial contact time provides for the ability to enact early countermeasures. A model is provided for the target sources, as well as metrics for computing costs associated with the model for reallocating robotic agents during an observation task. The effectiveness of the solution compared with an existing observation technique is demonstrated using simulation.

An underwater robotic assistant could help a human diver by illuminating work areas, fetching tools from the surface, or monitoring the diver's activity for abnormal behavior. However, in order for basic Underwater Human-Robot Interaction (UHRI) to be successful, the robotic assistant has to first be able to detect and track the diver. This paper discusses the detection and tracking of a diver with a high-frequency forward-looking sonar. The first step in the diver detection involves utilizing classical 2D image processing techniques to segment moving objects in the sonar image. The moving objects are then passed through a blob detection algorithm, and then the blob clusters are processed by the cluster classification process. Cluster classification is accomplished by matching observed cluster trajectories with trained Hidden Markov Models (HMM), which results in a cluster being classified as either a diver or clutter. Real-world results show that a moving diver can be autonomously distinguished from stationary objects in a noisy sonar image and tracked.

Home-based care modalities may yield many advantages in therapy and rehabilitation. Virtual systems for rehabilitation are an emerging technology that may be used to enhance the effectiveness of home-based care while simultaneously increasing the number of patients that physical therapists can provide care for. In this paper, we discuss a system that utilizes game design principles to develop a therapy game for upper extremity rehabilitation. We provide an overview of the system and show evidence based on assessments from adults and children using our virtual environment. Results indicate that the system we have developed can engage and encourage the target demographic by adhering to principles common in successful entertainment games.

Robotic assistance through telepresence technology is an emerging area in aiding the visually impaired. By integrating the robotic perception of a remote environment and transferring it to a human user through haptic environmental feedback, the disabled user can increase one's capability to interact with remote environments through the telepresence robot. This paper presents a framework that integrates visual perception from heterogeneous vision sensors and enables real-time interactive haptic represent-ation of the real world through a mobile manipulation robotic system. Specifically, a set of multi-disciplinary algorithms such as stereo-vision processes, three-dimensional map building algorithms, and virtual-proxy haptic rendering processes are integrated into a unified framework to accomplish the goal of real-world haptic exploration successfully. Results of our framework in an indoor environment are displayed, and its performances are analyzed. Quantitative results are provided along with qualitative results through a set of human subject testing. Our future work includes real-time haptic fusion of multi-modal environmental perception and more extensive human subject testing in a prolonged experimental design.

Inspired by large fatality rates due to fires in crowded areas and the increasing presence of robots in dangerous emergency situations, we have implemented a model of information propagation among evacuees. Information about the locations of exits and the relative confidence of the individual in the location of the exit disseminated through a simulated crowd of people during an evacuation modeled after The Station Nightclub fire of 2003. True believers were added to this system as individuals who refused to accept exit information from others, instead preferring to head to their own exit. This system was then tested to find what percentage of true believers most likely existed in the actual fire. Using this true believer percentage, robots were added to the environment to guide evacuees to the nearest exit. The number of people who believed a robot's instructions was varied to find what percentage of people need to trust these robots in order to exploit information propagation and thus increase survivability. As a lower bound, we have found that 30% of the evacuees should believe a robot's instructions to significantly increase survival rates.

Obesity is a growing health problem in the United States, especially among children. Indicators show that the rate of obesity for children age 12-19 years old has risen from 5% percent to 18% over the last ten years. To deal with the obesity epidemic, a number of technology interventions, including the use of robotics and virtual reality games, have arisen to motivate youth to become physically active. The difficulty though lies in providing a tool for health professionals to embed established clinical health protocols into these technologies. As such, in this paper we present a mixed reality system that translates physical demonstrations of various exercise protocols into movements for a robotic agent. This is accomplished by mapping real-time data from an RGB-D sensor to a robotic exercise coach. Details of the system are discussed and results from evaluation with 20 human subjects are provided.

In this paper, a low cost system for child interaction through turn taking and dance based on the Pleo robot platform is presented. This system is easily taught new dance movements through visual and haptic cues and provides immediate feedback of the learned motion, making it possible for individuals unfamiliar with robotics programming to alter its behavior through natural interaction.

In this paper, we discuss a pilot study that examines the effect technology-based learning activities have on engaging children with autism. We focus on the use of three activities that involve gaming and robotics. Preliminary data suggests that we can successfully teach children with autism using interactive learning scenarios while still maintaining the benefits found in traditional therapeutic playing scenarios. We give an overview of the learning activities in this paper, and provide results with respect to a case study involving a population of five children.

In this paper, we discuss an approach for enabling students with a visual impairment (VI) to validate the program sequence of a robotic system operating in the real world. We introduce a method that enables the person with VI to feel their robot's movement as well as the environment in which the robot is traveling. The design includes a human-robot interaction framework that utilizes multi-modal feedback to transfer the environmental perception to a human user with VI. Haptic feedback and auditory feedback are selected as primary methods for user interaction. Using this multi-modal sensory feedback approach, participants are taught to program their own robot to accomplish varying navigation tasks. We discuss and analyze the implementation of the method as deployed during two summer camps for middle-school students with visual impairment.

While the arctic possesses significant information of scientific value, surprisingly little work has focused on developing robotic systems to collect this data. For arctic robotic data collection to be a viable solution, a method for navigating in the arctic, and thus of assessing glacial terrain, must be developed. Segmenting the ground plane from the rest of the image is one common aspect of a visual hazard detection system. However, the properties of glacial images, namely low contrast, overcast sky, and cloud, mountain, and snow sharing common colors, pose difficulties for most visual algorithms. A horizon line detection scheme is presented which uses multiple visual cues to rank candidate horizon segments, then constructs a horizon line consistent with those cues. Weak cues serve to reinforce a selected path, while strong cues have the ability to redirect it. Further, the system infers the horizon location in areas that are visually ambiguous. The performance of the proposed system has been tested on multiple data sets collected on two different glaciers in Alaska, and compares favorably, both in terms of time and classification performance, to representative segmentation algorithms from several different classes.

Evacuating a building in an emergency situation can be very confusing and dangerous. Exit signs are static and thus have no ability to convey information about congestion or danger between the sign and the actual exit door. Emergency personnel may arrive too late to assist in an evacuation. Robots, however, can be stored inside of buildings and can be used to guide evacuees to the best available exit. To enable this process, evacuation robots must have an understanding of how people react in emergency situations. By incorporating a model of human panic behavior, these robots can effectively guide crowds of people to zones of safety. In this paper, we discuss an initial design of these robots and their behaviors. Preliminary simulation results show that a significantly larger proportion of people are evacuated with robot assistance than without.

This paper presents preliminary results of a mobile manipulator robot tasked to play the classic Towers of Hanoi game. We first discuss the control algorithms necessary to enable necessary game-playing behavior and provide results of implementing our methodology in a high fidelity 3D environment. After attaining success in the simulation environment, we provide results on implementation of the same control software using physical robot hardware. Additionally, preliminary analysis for implementing analog Proportional-Integral-Derivative (PID) control on this platform using a floating-gate based reconfigurable analog IC is explored. Using this concept of floating gate analog arrays for control enables off-loading of the processing, which could be helpful for real-time implementation of robot behavior.

With the advent of new technologies, robotic surveying systems are being developed to facilitate the collection of ground-based information to validate and complement data collected by traditional and satellite-based instruments. The development of such systems necessitates an accurate set of reference data. Given the limitations of current in situ measurement methods to aid remote sensing, this paper outlines a method for the creation of three-dimensional data from the most common public data source, 2D contour maps. Using image processing and interpolation techniques, this method was first tested against data collected by a robotic survey system and against methods that a human expert would use. Comparatively, our method yielded vertical RMSE in the range of (0.006066 - 0.39) [m] for different horizontal spatial resolutions. Twenty additional sample contour maps were identified to further vet our method against that of a human expert as a function of the 3D interpolation method selected. These tests provided errors in the centimeter range and also revealed that the linear triangular mesh interpolator is the best choice for this type of image input data.

Efficiently measuring environmental phenomena (e.g., elevation, chemical composition, and mineral density) is a task typically reserved for the geoscience community. Recent robotic systems with the potential for addressing the task of sampling currently exist, yet their navigation strategies (and subsequently sampling strategies) are seldom a function of the spatial change in the measured phenomena of interest. Solutions are especially void for intelligent systems to which resource constraints are applied (i.e., battery power and experimentation time) while complete coverage of an area is expected. In this paper, we discuss the implementation of a custom navigation strategy based on immediately-sensed data that, when combined with spatial interpolation techniques, yields a re-creation of the surveyed space with root mean squared error that meets accepted mapping standards. Our methodology employs an adaptive coverage algorithm which succeeds in lowering the RMS error when compared to other navigation techniques. Our results are validated in simulation by considering: 1) randomly-generated terrains and 2) realistic digital elevation map (DEM) data transposed from publically available terrain contour maps.

In this paper, we discuss a methodology to build a system for a robot playmate that extracts and sequences low-level play primitives during a robot-child interaction scenario. The motivation is to provide a robot with basic knowledge of how to manipulate toys in an equivalent manner as a human does - as a first step in engaging children in cooperative play. Our approach involves the extraction of play primitives based on observation of motion gradient vectors computed from the image sequence. Hidden Markov Models (HMMs) are then used to recognize 14 different play primitives during play. Experimental results from a data set of 100 play scenarios including child subjects demonstrate 86.88% accuracy recognizing and sequencing the play primitives.

In this paper, we discuss a coordinated haptic training architecture useful for transferring expertise in teleoperation-based manipulation between two human users. The objective is to construct a reality-based haptic interaction system for knowledge transfer by linking an expert's skill with robotic movement in real time. The benefits from this approach include 1) a representation of an expert's knowledge into a more compact and general form by learning from a minimized set of training samples, and 2) an increase in the capability of a novice user by coupling learned skills absorbed by a robotic system with haptic feedback. In order to evaluate our ideas and present the effectiveness of our paradigm, human handwriting is selected as our experiment of interest. For the learning algorithms, artificial neural network (ANN) and support vector machine (SVM) are utilized and their performances are compared. For the evaluation of the performance of the output of the learning modules, a modified Longest Common Subsequence (LCSS) algorithm is implemented. Results show that one or two experts' samples are sufficient for the generation of haptic training knowledge, which can successfully recreate manipulation motion with a robotic system and transfer haptic forces to an untrained user with a haptic device. Also in the case of handwriting comparison, the similarity measures result in up to an 88% match even with a minimized set of training samples.

The ability of robotic units to autonomously navigate various terrains is critical to the advancement of robotic operation in natural environments. Next generation robots will need to adapt to their environment in order to accomplish tasks that are either too hazardous, too time consuming, or physically impossible for human-beings. Such tasks may include accurate and rapid explorations of various planets or potentially dangerous areas on Earth. Furthermore, because terrain variability typically increases as the distance that a rover traverses increases, it will be beneficial for robotic units to adapt to their surroundings. As a result, this research investigates a navigation control methodology for a multi-modal locomotive robot based upon passive perception. Surface estimation for robot reconfigurability is implemented using a region growing method, which characterizes the traversability of the terrain, in conjunction with passive perception regarding motion. A mathematical approach is then implemented that inherits human psychological aspects to direct necessary navigation behavior to control robot mobility. Physical experimentations in a simulated Mars yard are presented to validate the methodology.

Heterogeneous multi-robot networks require novel tools for applications that require achieving and maintaining formations. This is the case for distributing sensing devices with heterogeneous mobile sensor networks. Here, we consider a heterogeneous multi-robot network of mobile robots. The robots have a limited range in which they can estimate the relative position of other network members. The network is also heterogeneous in that only a subset of robots have localization ability. We develop a method for automatically configuring the heterogeneous network to deploy a desired formation at a desired location. This method guarantees that network members without localization are deployed to the correct location in the environment for the sensor placement.

When young children play, they often manipulate toys that have been specifically designed to accommodate and stimulate their perceptual-motor skills. Robotic playmates capable of physically manipulating toys have the potential to engage children in therapeutic play and augment the beneficial interactions provided by overtaxed care givers and costly therapists. To date, assistive robots for children have almost exclusively focused on social interactions and teleoperative control. Within this paper we present progress towards the creation of robots that can engage children in manipulative play. First, we present results from a survey of popular toys for children under the age of 2 which indicates that these toys share simplified appearance properties and are designed to support a relatively small set of coarse manipulation behaviors. We then present a robotic control system that autonomously manipulates several toys by taking advantage of this consistent structure. Finally, we show results from an integrated robotic system that imitates visually observed toy playing activities and is suggestive of opportunities for robots that play with toys.

In this paper we extend our previous work where the mean of the global cost was used as a performance metric for distributed task allocation algorithms. In this case, we move a step forward and calculate the variance of the global cost. This second parameter gives us a better understanding of the distributed algorithm performance, i.e., we can estimate how much the algorithm behavior diverts from its mean. The normal distribution, computed from the theoretical mean and variance, is shown to be suitable for modeling the global cost. This approximation enables us to compare our algorithm theoretically in different cases.

Learning control strategies from examples has been identified as an important capability for many robotic systems. In this work we show how the learning process can be aided by autonomously filtering the training set provided to improve key properties of the learning process. Demonstrated with data gathered for manipulation tasks, the results herein show the improved performance when autonomous filtering is applied. The filtration method, with no prior knowledge of the task, was able to partition the training sets into sets almost equal to expertly labeled sets. In the case where the filter did not produce the same groupings as the expert user, the method still permitted a controller to be trained which demonstrated a success rate of 92%.

In ground-based military maneuvers, group formations require flexibility when traversing from one point to the next. For a human-led team of semi-autonomous agents, a certain level of awareness demonstrated by the agents regarding the quality of the formation is preferable. Through the use of a Multi-Robot System (MRS), this work combines leader-follower principles augmented by an assistive formation maintenance (AFM) method to improve formation keeping and demonstrate a formation-in-motion concept. This is achieved using the Robot Mean Task Allocation method (RTMA), a strategy used to allocate formation positions to each unit within a continuously mobile MRS. The end goal is to provide a military application that allows a soldier to efficiently tele-operate a semi-autonomous MRS capable of holding formation amidst a cluttered environment. Baseline simulation is performed in Player/Stage to show the applicability of our developed model and its potential for expansive research.

Novel tools are needed to deploy multi-agent networks in applications that require a high degree of accuracy in the achievement and maintenance of geometric formations. This is the case when deploying distributed sensing devices across large spatial domains. Through so-called embedded graph grammars (EGGs), this paper develops a method for automatically generating control programs that ensure that a multi-robot network is deployed according to the desired configuration. This paper presents a communication protocol needed for implementing and executing the control programs in an accurate and deadlock-free manner.

Arctic regions present one of the harshest environments on earth for people or mobile robots, yet many important scientific studies, particularly those involving climate change, require measurements from these areas. For the successful deployment of mobile sensors in the arctic, a reliable, fault tolerant, low-cost method of navigating must be developed. One aspect of an autonomous navigation system must be an assessment of the local terrain, including the slope of nearby regions. Presented here is a method of estimating the slope of the terrain in the robot's coordinate frame using only a single camera, which has been applied to both simulated arctic terrain and real images. The slope estimates are then converted into the global coordinate frame using information from a roll sensor, used as an input to a fuzzy logic navigation scheme, and tested in a simulated arctic environment.

In this paper, we discuss a methodology to extract play primitives, defined as a sequence of low-level motion behaviors identified during a playing action, such as stacking or inserting a toy. Our premise is that if a robot could interpret the basic movements of a humanpsilas play, it will be able to interact with many different kinds of toys, in conjunction with its human playmate. As such, we present a method that combines motion behavior analysis and behavior sequencing, which capitalizes on the inherent characteristics found in the dynamics of play such as the limited domain of the objects and manipulation skills required. In this paper, we give details on the approach and present results from applying the methodology to a number of play scenarios.

Applications of robotics are quickly changing. Just as computer use evolved from research purposes to everyday functions, applications of robotics are making a transition to mainstream usage. With this change in applications comes a change in the user base of robotics, and there is a pronounced move to reduce the complexity of robotic control. The move to reduce complexity is linked to the separation of the role of robot designer and robot operator. For many target applications, the operator of the robot needs to be able to correct and augment its capabilities. One method to enable this is learning from human data, which has already been successfully applied to robotics. We assert that this learning process is only viable when the demonstrated human behavior is coherent. In this work we test the hypothesis that quantifying the coherence in the provided instruction can provide useful information about the progress of the learning process. We discuss results from the application of this method to reactive behaviors. Such behaviors permit the learning process to be computationally tractable in real-time. These results support the hypothesis that coherence is important for this type of learning and also show that this property can be used to provide an avenue for self regulation of the learning process.

In this paper, we discuss a methodology that employs vision-based force guidance techniques for improving human performance with respect to a teleoperated manipulation system. The primary focus of the approach is to study the effectiveness of guidance forces in a haptic system to enable ease-of-use for human operators performing common manipulation activities necessary for achievement of everyday tasks. By designing force feedback signals constructed only from visual imagery data as input into a haptic device, we show the impact on human performance during the teleoperation sequence. The methodology is explained in detail, and results of implementation on object-centering and object-approaching tasks with our divided force guidance approach are presented.

In this paper, an analysis of a market-based approach applied to the initial formation problem is presented. This problem tries to determine which mobile sensor should go to each position of a desired formation in order to minimize an objective. In our case, this objective is the global distance traveled by all the mobile sensors. In this analysis, a bound on the efficiency for the market-based algorithm is calculated and it is shown that the relative difference as compared with the optimal solution increases with the logarithm of the total number of mobile sensors. The theoretical results are validated with numerous simulations.

In this paper, we present a hierarchical methodology that learns new walking gaits autonomously while operating in an uncharted environment, such as on the Mars planetary surface or in the remote Antarctica environment. The focus is to maintain persistent forward locomotion along the body axis, while navigating in natural terrain environments. The hierarchical strategy consists of a finite state machine that models the state of leg orientations coupled with a modified evolutionary algorithm to learn necessary leg movement sequences. Locomotion behavior is assessed by monitoring the robot's progress toward a specified goal location. Details of the methodology are discussed, and experimental results with a six-legged robot are presented.

In this paper we discuss a method for transferring human knowledge to a robotic platform via teleoperation. The method combines unsupervised clustering and classification with interactive instruction to enable behavior capture in a transferable form. We discuss the approach in both simulation and robotic hardware platform to show the capability of the learning system. In this work we also present a definition and associated metric for trustworthiness, and relate this quantity to system performance. Improved performance and trustworthiness are motivations for our application of interactive learning, and we present results that indicate that these were indeed attained.

This paper discusses an approach for predicting system performance resulting from humans and robots performing repetitive tasks in a collaborative manner. The methodology uses a systematic approach that incorporates the various effects of workload on human performance, and estimates resulting performance attributes derived between teleoperated and autonomous control of robotic systems. Performance is determined by incorporating capabilities of the human and robotic agent based on accomplishment of functional operations and effect of cognitive stress due to continuous operation by the human agent. This paper provides an overview of the prediction system and discusses its implementation on a simulated rendezvous/docking task.

In this paper, we focus on the problem of maximizing system performance for future space exploration missions involving both human and robot agents. One of the main challenges in human-robot interaction scenarios is determining which tasks are best done with either human, robotic systems, or in collaboration with each. Such partitioning of the task space must acknowledge the capabilities of both agents, as well as incorporate the effect of repetitive workload, or stress, on the human operator. Our methodology for role allocation, which typically consists of either the human or the machine executing a single task, is based on predicting system performance of a given scenario by incorporating the concept of task switching. Task switching is defined as the process of alternating or switching attention between tasks when responding to a sequence of stimulus presentations. Using this concept, system performance can be predicted and used to determine an optimal allocation of tasks to be divided between human controlled and autonomous robotic systems to minimize mental workload while maximizing task performance. We provide details of the approach in this paper and present our results as applied to a simulated rendezvous/docking mission scenario

In this paper, we present a methodology to assess system performance of human-robotic systems in achievement of collective tasks such as habitat construction, geological sampling, and space exploration. The methodology uses a systematic approach that assesses performance by incorporating capabilities of both human and robotic agents based on accomplishment of functional operations and effect of cognitive stress due to continuous operation by the human agent. In this paper, we provide an overview of the assessment system and discuss its implementation on a representative habitat construction task.

This paper discusses a virtual environment that enables human agents to develop the skills necessary to control a mobile robot through the implementation of exploratory learning practices. The interface connects the human user to both a virtual and physical robot resident in the real-world, and allows evaluation of human performance using a framework that analyzes execution parameters during human operation. The execution data is then used to compare the capability of human agents to learn the skill sets necessary to control the robot during a novel task situation. We give an overview of the environment, as well as the experimental results comparing the performance of multiple operators learning to control a virtual robot.

In this paper, we present an approach to transfer human expertise for learning off-road navigation behavior to an autonomous mobile robot. The methodology uses the concept of humanized intelligence to combine principal component analysis and neural network learning to embed human driving expertise onto mobile robots. The algorithms are tested in the field using a commercial Pioneer 2AT robot to demonstrate autonomous traversal over rough natural terrain.

In this paper, we propose a method for multi-robot task allocation based on the concept of task decomposition for a lunar mission scenario. This methodology focuses on segmenting a task scenario into a sequence of operations called functional primitives that are defined a priori by a set of performance metrics and resource requirements. In real-time, multiple robotic agents determine their capabilities and skill sets associated with the defined functional primitives in order to determine a suitable allocation scheme. We discuss the methodology in detail and provide results for a simulated lunar mission construction scenario using the Multi-Agent Robot Simulator for Lunar Construction (MARS-LC) system.

This paper presents a new method for integrating terrain maps into a reactive navigation strategy of field mobile robots operating on rough terrain. The method incorporates the Regional Traversability Map, a fuzzy map representation of traversal difficulty of the regional terrain, into the navigation logic. A map-based regional navigation behavior provides speed and direction recommendations based on the current status of the robot. In addition, recommendations from two sensor-based reactive behaviors, local avoid-obstacle and regional traverse-terrain, are fused with the map-based regional behavior to construct a comprehensive navigation system. The algorithms are tested both in graphical simulations and in the field using a commercial Pioneer 2AT robot to demonstrate traversal over rough natural terrain.

This paper presents a novel information fusion methodology for intelligent terrain analysis. In our application, we define information as terrain characteristics derived from sensor data extracted from on-board spacecraft sensors. The fuzzy-logic construct allows one to represent the terrain characteristics using an easily understandable, linguistic approach. Once derived, these fuzzy terrain characteristics are blended together using a fuzzy rule base to produce a coherent representation of terrain safety. The fused information is then used to autonomously select a safe landing site for spacecraft touchdown. The fuzzy terrain analysis and fusion methodology is explained in detail. Computer simulation results are provided to show the viability of the approach.

Presents a fuzzy terrain-based path planning method for planetary rovers operating on rough natural terrain. The focus of this approach is on planning an optimally safe path of minimum traversal cost, which is calculated from linguistic descriptors of terrain traversability. The method incorporates the traversability map, a fuzzy map representation of traversal difficulty of the terrain, into the path planning logic. The search methodology uses a traversal cost function that is derived directly from this traversability map. The path planning method is developed in detail and experimental results are presented.

Operational safety and health monitoring are critical matters for autonomous field mobile robots such as planetary rovers operating on challenging terrain. This paper describes relevant rover safety and health issues and presents an approach to maintaining vehicle safety in a mobility and navigation context. The proposed rover safety module is composed of two distinct components: safe attitude (pitch and roll) management and safe traction management. Fuzzy logic approaches to reasoning about safe attitude and traction management are presented, wherein inertial sensing of safety status and vision–based neural network perception of terrain quality are used to infer safe speeds of traversal. Results of initial field tests and laboratory experiments are also described. The approach provides an intrinsic safety cognizance and a capacity for reactive mitigation of robot mobility and navigation risks.

This paper presents a new strategy for behavior-based navigation of field mobile robots on challenging terrain, using a fuzzy logic approach and a novel measure of terrain traversability. A key feature of the proposed approach is real-time assessment of terrain characteristics and incorporation of this information in the robot navigation strategy. Three terrain characteristics that strongly affect its traversability, namely, roughness, slope, and discontinuity, are extracted from video images obtained by on-board cameras. This traversability data is used to infer, in real time, the terrain Fuzzy Rule-Based Traversability Index, which succinctly quantifies the ease of traversal of the regional terrain by the mobile robot. A new traverse-terrain behavior is introduced that uses the regional traversability index to guide the robot to the safest and the most traversable terrain region. The regional traverse-terrain behavior is complemented by two other behaviors, local avoid-obstacle and global seek-goal. The recommendations of these three behaviors are integrated through adjustable weighting factors to generate the final motion command for the robot. The weighting factors are adjusted automatically, based on the situational context of the robot. The terrain assessment and robot navigation algorithms Are implemented on a Pioneer commercial robot and field-test studies are conducted. These studies demonstrate that the robot possesses intelligent decision-making capabilities that are brought to bear in negotiating hazardous terrain conditions during the robot motion.

Sun sensors are widely used in spacecraft attitude determination subsystems to provide a measurement of the Sun vector in spacecraft coordinates. At the Jet Propulsion Laboratory, California Institute of Technology, there is an ongoing research activity to utilize Micro Electro Mechanical Systems (MEMS) processes to develop a smaller and lighter Sun sensor for space applications. A prototype Sun sensor has been designed and constructed. It consists of a piece of silicon coated with a thin layer of chrome, and a layer of gold with hundreds of small pinholes, placed on top of an image detector at a distance of less than a millimeter. Images of the Sun are formed on the detector when the Sun illuminates the assembly. Software algorithms must be able to identify the individual pinholes on the image detector and calculate the angle to the Sun. Fuzzy image processing is utilized in this process. This paper describes how the fuzzy image processing is implemented in the instrument. Also, a camera pin hole model is constructed and used to evaluate the accuracy of the Sun sensor.

This paper presents a rule-based fuzzy traversability index that quantifies the ease-of-traversal of a terrain by a mobile robot based on real-time measurements of terrain characteristics retrieved from imagery data. These characteristics include, but are not limited to slope, roughness, hardness, and discontinuity. The proposed representation of terrain traversability incorporates an intuitive, linguistic approach for expressing terrain characteristics that is robust with respect to imprecision and uncertainty in the terrain measurements. The terrain assessment method is tested and validated with a set of real-world imagery data. These tests demonstrate the capability of the terrain classification algorithm for perceiving hazards associated with terrain traversal.

Presents a strategy for autonomous navigation of field mobile robots on hazardous natural terrain using a fuzzy logic approach and a measure of terrain traversability. The navigation strategy comprises three simple, independent behaviors: seek-goal, traverse-terrain, and avoid-obstacle. The recommendations from these three behaviors are combined through appropriate weighting factors to generate the final steering and speed commands that are executed by the robot. The weighting factors are produced by fuzzy logic rules that take into account the current status of the robot. This navigation strategy requires no a priori information about the environment, and uses the on-board traversability analysis to enable the robot to select relatively easy-to-traverse paths autonomously. Field test results obtained from implementation of the proposed algorithms on the commercial Pioneer All Terrain rover are presented. These results demonstrate the real-time capabilities of the terrain assessment and fuzzy logic navigation algorithms.

Operational safety and health monitoring are critical matters for autonomous field mobile robots such as planetary rovers operating on challenging terrain. The paper describes relevant rover safety and health issues and presents an approach to maintaining vehicle safety in a navigational context. The proposed rover safety module is composed of two distinct components: safe attitude (pitch and roll) management and safe traction management. Fuzzy logic approaches to reasoning about safe attitude and traction management are presented, wherein sensing of safety status and perception of terrain quality are used to infer safe speeds of traversal. Results of field tests and laboratory experiments are also described. The approach provides an intrinsic safety cognizance and a capacity for reactive mitigation of navigation risks.

This paper presents a novel technique for real-time measurement of terrain characteristics and incorporation of this information into the navigation strategy of an autonomous mobile robot. The proposed methodology utilizes a fuzzy logic framework for on-board analysis of terrain traversability, and develops a set of fuzzy navigation rules that guide the rover toward the safest and the most traversable terrain. In addition, a simple goal-seeking behavior is used to drive the rover from its initial position to a user-specified goal position. The overall navigation strategy, consisting of terrain-traverse and goal-seeking behaviors, requires no a priori information about the environment, and uses the on-board traversability analysis to enable the rover to select easy-to-traverse paths to the goal autonomously. The terrain traversability navigation rules are tested and validated with a set of physical rover experiments. These experiments demonstrate the real-time capability of the terrain assessment and fuzzy navigation algorithms.