Vaibhav Unhelkar
Date:
Speaker
Vaibhav Unhelkar is an Assistant Professor at Rice University, where he leads the Human-Centered AI & Robotics group. His work spans the development of robotic assistants, intelligent tutors, and decision-support systems aimed at enhancing human performance in domains ranging from healthcare to disaster response. Underpinning these systems are Unhelkar’s contributions to imitation learning and explainable AI, designed to train robots, humans, and human-AI teams. He earned his Ph.D. in Autonomous Systems from MIT in 2020 and completed his undergraduate studies at IIT Bombay in 2012. Unhelkar is actively engaged in the AI and robotics research communities, serving on the editorial board of robotics and AI conferences. His research and service have been recognized with awards from the International Foundation for Autonomous Agents and Multi-Agent Systems (AAMAS), among others.
Abstract
We are steadily approaching a future where humans work with robotic assistants, teammates, and even tutors. While significant effort is being dedicated to training robots to work with humans, much less attention has been given to the human side of this partnership. For instance, we see increasingly capable AI models and algorithms that enable robots to infer human intent, plan interactions, and communicate with them. However, human users often lack understanding of how robots will react in novel situations and receive minimal training to work with robots. To ensure the safe and effective use of robots, I argue that training human users is equally critical. This talk will present human-centered computing methods for generating and delivering such training. First, I will discuss policy summarization methods that can help generate the training content, enabling humans to develop a Theory of Mind regarding the robot. Next, I will discuss mechanisms for delivering this training. The talk will conclude with an overview of open problems in making this training more effective through personalization and interactions.
Papers Covered
- Unhelkar, Vaibhav, Shen Li, and Julie Shah. “Decision-making for bidirectional communication in sequential human-robot collaborative tasks.” Proceedings of the ACM/IEEE International Conference on Human-Robot Interaction. 2020.
- Rong, Yao, Peizhu Qian, Vaibhav Unhelkar, and Enkelejda Kasneci. “I-CEE: tailoring explanations of image classification models to user expertise.” In Proceedings of the AAAI Conference on Artificial Intelligence, vol. 38, no. 19, pp. 21545-21553. 2024.
- Qian, Peizhu, and Vaibhav Unhelkar. “Evaluating the role of interactivity on improving transparency in autonomous agents.” Companion of the ACM/IEEE International Conference on Human-Robot Interaction. 2024.
- Qian, Peizhu, Harrison Huang, and Vaibhav Unhelkar. “PPS: Personalized Policy Summarization for Explaining Sequential Behavior of Autonomous Agents.” Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society. Vol. 7. 2024.
- Qian, Peizhu, Filip Bajraktari, Carlos Quintero-Peña, Qingxi Meng, Shannan Hamlin, Lydia Kavraki, and Vaibhav Unhelkar. “ASTRID: A Robotic Tutor for Nurse Training to Reduce Healthcare-Associated Infections.” Robotics: Science and Systems. 2025.