Browsing by Author "Ghafurian, Moojan"

Now showing 1 - 5 of 5

Exploring Human Teachers' Interpretations of Trainee Robots' Nonverbal Behaviour and Errors
(University of Waterloo, 2021-04-22) Aliasghari, Pourya; Dautenhahn, Kerstin; Nehaniv, Chrystopher L.; Ghafurian, Moojan
In the near future, socially intelligent robots that can learn new tasks from humans may become widely available and gain an opportunity to help people more and more. In order to successfully play a role, not only should intelligent robots be able to interact effectively with humans while they are being taught, but also humans should have the assurance to trust these robots after teaching them how to perform tasks. When human students learn, they usually provide nonverbal cues to display their understanding of and interest in the material. For example, they sometimes nod, make eye contact or show meaningful facial expressions. Likewise, a humanoid robot's nonverbal social cues may enhance the learning process, in case the provided cues are legible for human teachers. To inform designing such nonverbal interaction techniques for intelligent robots, our first work investigates humans' interpretations of nonverbal cues provided by a trainee robot. Through an online experiment (with 167 participants), we examine how different gaze patterns and arm movements with various speeds and different kinds of pauses, displayed by a student robot when practising a physical task, impact teachers' understandings of the robot’s attributes. We show that a robot can appear differently in terms of its confidence, proficiency, eagerness to learn, etc., by systematically adjusting those nonverbal factors. Human students sometimes make mistakes while practising a task, but teachers may be forgiving about them. Intelligent robots are machines, and therefore, they may behave erroneously in certain situations. Our second study examines if human teachers for a robot overlook its small mistakes made when practising a recently taught task, in case the robot has already shown significant improvements. By means of an online rating experiment (with 173 participants), we first determine how severe a robot’s errors in a household task (i.e., preparing food) are perceived. We then use that information to design and conduct another experiment (with 139 participants) in which participants are given the experience of teaching trainee robots. According to our results, perceptions of teachers improve as the robots get better in performing the task. We also show that while bigger errors have a greater negative impact on human teachers' trust compared with the smaller ones, even a small error can significantly destroy trust in a trainee robot. This effect is also correlated with the personality traits of participants. The present work contributes by extending HRI knowledge concerning human teachers’ understandings of robots, in a specific teaching scenario when teachers are observing behaviours that have the primary goal of accomplishing a physical task.
Exploring the Use of Social Robots in Promoting University Students’ Mental Well-Being
(University of Waterloo, 2023-12-12) Rasouli, Samira; Dautenhahn, Kerstin; Ghafurian, Moojan
Mental health problems within university student populations are a growing concern. In recent years, university students have reported increased symptoms of stress and anxiety, which can negatively impact their mental well-being and academic performance. However, many students do not seek or receive support for these challenges. With recent technological advances in social robots and their expanding capabilities, there is potential for social robots to extend and complement mental health care interventions and enhance access to care. This thesis introduces a novel application of social robots in robot-assisted interventions for addressing stress and anxiety in social situations. As an initial step to developing effective technology-based interventions, it is essential to identify design elements and functionalities that are perceived as engaging and useful by students. Therefore, our first study explores students’ perceptions of, and preferences for, using different types of intelligent agents (e.g., virtual agents, social robots, etc.) to support their mental well-being, specifically to cope with feelings of stress and anxiety in social situations typically encountered within a university context (e.g., engaging in a group discussion, delivering presentations, expressing opinions, etc.). Two online surveys were conducted: a pilot study on Amazon Mechanical Turk (N = 85) and a larger study at the University of Waterloo (N = 1054). The results provide insights into different design elements (e.g., confidentiality, accessibility and enjoyment) as well as social and technical capabilities (e.g., ability to understand conversation) to consider when designing intelligent agents to help address stress and anxiety among university students. The results also showed that public speaking is one of the most anxiety-provoking social situations at university, and animal-like robots were the most selected agents among the given choices of different types of intelligent agents to use for managing anxiety related to the students' most anxiety-provoking social situations. Participants also endorsed a variety of activities that they thought could be beneficial for managing anxiety related to their most anxiety-provoking situation, including receiving personalized coaching for activities and tasks that support mental well-being (e.g., meditation and yoga), getting help with thinking calming or positive thoughts, and getting help with building confidence, for which they indicated to be willing to use an intelligent agent. In the next step, we used the findings of our first study to design a robotic mental well-being coach aimed at assisting university students in managing public speaking anxiety. To this end, in our second study, we conducted collaborative co-design sessions with five mental health professionals (the envisioned secondary users) to identify the design-related needs, including robot behaviours and interactions. We employed a co-design approach to ensure that the design of the robot's behaviours and functionalities is grounded in evidence-based practices and aligns with the best practices of psychological support in this application area that can be delivered by a robotic coach. Furthermore, we carried out a participatory study involving six university students (the envisioned primary users) to gather their opinions for further improvements to the robotic coach. Students provided feedback on the behaviours and functionality of the robot and generally found the robot engaging, relaxing, knowledgeable, and beneficial for learning relaxation exercises. Lastly, after implementing final improvements to the behaviour and functionalities of the robotic coach based on feedback from students and with the assistance of mental health professionals, we conducted our third study with university students (N=50) to evaluate the usability and acceptability of the robotic mental well-being coach designed in the second study to help university students manage their public speaking anxiety. The findings demonstrated a usability score of 84.05 for the robotic coach and high acceptability among university students who perceived the robot as competent and knowledgeable. In addition, a significant improvement in participants' self-reported moods was observed following the study. The qualitative analysis suggests that university students had a positive impression of the robotic mental well-being coach and their interaction. Overall, the findings of this study are encouraging and indicate the potential of social robots as mental well-being coaches to support university students' public speaking anxiety. Future research should investigate the users' experiences and the effectiveness of the robotic mental well-being coach in assisting university students to cope with public speaking anxiety in the long term over multiple interaction sessions.
How Do Different Modes of Verbal Expressiveness of a Student Robot Making Errors Impact Human Teachers' Intention to Use the Robot?
(Association for Computing Machinery, 2021-11-09) Aliasghari, Pourya; Ghafurian, Moojan; Nehaniv, Chrystopher L.; Dautenhahn, Kerstin
When humans make a mistake, they often try to employ some strategies to manage the situation and possibly mitigate the negative effects of the mistake. Robots that operate in the real world will also make errors and therefore might benefit from such recovery strategies. In this work, we studied how different verbal expression strategies of a trainee humanoid robot when committing an error after learning a task influence participants’ intention to use it. We performed a virtual experiment in which the expression modes of the robot were as follows: (1) being silent; (2) verbal expression but ignoring any errors; or (3) verbal expression while mentioning any error by apologizing, as well as acknowledging and justifying the error. To simulate teaching, participants remotely demonstrated their preferences to the robot in a series of food preparation tasks; however, at the very end of the teaching session, the robot made an error (in two of the three experimental conditions). Based on data collected from 176 participants, we observed that, compared to the mode where the robot remained silent, both modes where the robot utilized verbal expression could significantly enhance participants' intention to use the robot in the future if it made an error in the last practice round. When no error occurred at the end of the practice rounds, a silent robot was preferred and increased participants' intention to use.
Integrating Affective Expressions into Robot-Assisted Search and Rescue to Improve Human-Robot Communication
(University of Waterloo, 2021-09-13) Akgun, Sami Alperen; Dautenhahn, Kerstin; Ghafurian, Moojan; Crowley, Mark
Unexplained or ambiguous behaviours of rescue robots can lead to inefficient collaborations between humans and robots in robot-assisted SAR teams. To date, rescue robots do not have the ability to interact with humans on a social level, which is believed to be an essential ability that can improve the quality of interactions. This thesis research proposes to bring affective robot expressions into the SAR context to provide rescue robots social capabilities. The first experiment presented in Chapter 3 investigates whether there is consensus in mapping emotions to messages/situations in Urban Search and Rescue (USAR) scenarios, where efficiency and effectiveness of interactions are crucial to success. We studied mappings between 10 specific messages, presented in two different communication styles, reflecting common situations that might happen during search and rescue missions and the emotions exhibited by robots in those situations. The data was obtained through a Mechanical Turk study with 78 participants. The findings support the feasibility of using emotions as an additional communication channel to improve multi-modal human-robot interaction for urban search and rescue robots and suggest that these mappings are robust, i.e., are not affected by the robot’s communication style. The second experiment was conducted on Amazon Mechanical Turk as well with 223 participants. We used Affect Control Theory (ACT) as a method for deriving the mappings between situations and emotions (similar to the ones in the first experiment) and as an alternative method to obtaining mappings that can be adjusted for different emotion sets (Chapter 4). The results suggested that there is consistency in the choice of emotions for a robot to show in different situations between the two methods used in the first and second experiment, indicating the feasibility of using emotions as an additional modality in SAR robots. After validating the feasibility of bringing emotions to SAR context based on the findings from the first two experiments, we created affective expressions based on Evaluation, Potency and Activity (EPA) dimensions of ACT with the help of LED lights on a rescue robot called Husky. We evaluated the effect of emotions on rescue workers’ situational awareness through an online Amazon Mechanical Turk Study with 151 participants (Chapter 5). Findings indicated that participants who saw Husky with affective expressions (conveyed through lights) had better perception accuracy of the situation happening in the disaster scene than participants who saw the videos of the Husky robot without any affective lights. In other words, Husky with affective lights improved participants’ situational awareness.
Pavement Performance Prediction Using Machine Learning and Instrumentation in Smart Pavement
(University of Waterloo, 2022-09-19) Kang, Jianqi; Ghafurian, Moojan; Baaj, Hassan
The optimization of pavement Maintenance and Rehabilitation (M&R) planning and costs has been historically proven as a complex task. In recent years, Artificial Intelligence (AI) and Machine Learning (ML) applications in pavement engineering data analytics have been gaining momentum. These advanced techniques have shown promising results in civil engineering and transportation asset management. Therefore, designing a smart pavement framework that relies on the actual pavement responses and in-service condition can help with utilising the ML approach toward better understanding the performance of pavements. To implement the concept of “Smart Pavement”, constructing an interactive pavement pilot section that provides the necessary data feedback to improve the decision-makings of M&R would be needed. This thesis focuses on some aspects of the design of in-situ pavement monitoring and the applying selected machine learning techniques for pavement performance prediction. In order to design an effective pavement instrumentation plan, a literature review was conducted to identify and evaluate the major in-situ monitoring devices and previous case studies. Innovative technologies of Structural Health Monitoring (SHM) were also discussed as a part of the sensory system. A potential pilot section was identified by the Region of Waterloo, for which the pavement structure and technical details were retrieved. Based on the results from the literature review and the evaluation of the proposed section details, a preliminary instrumentation layout has been proposed. Next, the interaction between the proposed embedded sensors and surrounding pavement structure under traffic loading was further studied to evaluate the effect of pavement instrumentation on actual structural responses. Therefore, a series of finite element analysis (FEA) scenarios were defined, and modelling was conducted using ABAQUS to quantify the artefact impacts of the sensors on the pavement responses. Based on the FEA results, high stress- and strain-concentration areas were located which can be used to optimize the design of sensor layout, leading to capturing representative critical pavement responses. Consequently, sensor spacing criteria were suggested to avoid device interference for the response measurement. Furthermore, it would be informative to know how, and which AI/ML methods have been previously used for pavement performance prediction purposes. A systematic literature review iii was conducted indicating that majority of studies used Artificial Neural Network (ANN) of which the prediction process is unexplainable to predict International Roughness Index (IRI) resulting in high prediction accuracies (R2 >= 0.9). A Decision Tree (DT) model and a Random Forest (RF) model were developed using the most commonly used input data retrieved from the Long-Term Pavement Performance (LTPP) database to predict IRI. Finally, after the pruning process, the DT model and RF model resulted in a cross-validation accuracy (R^2) of 0.846 and 0.859, respectively. The single tree from the DT model is less complex than the trees from the RF model. Further studies on machine learning model development should be conducted to refine prediction accuracy. Finally, recommendation for future data collection standards from pilot sections were provided to help with developing a pavement response database that can overcome the inconsistencies in the existing LTPP database and potentially improve the reliability of the future pavement performance modelling.