Suresh Kumaar Jayaraman
Robotics Institute . Carnegie Mellon University . Postdoctoral researcher at TBD, HARP, and RASL
My research aims to develop autonomous systems that can seamlessly integrate into complex, human-centric environments. Motivated by the critical need for mutual understanding between humans and autonomous systems, my work focuses on two key thrusts:
- Modeling human behavior to enable robots to interpret human intentions, actions, and mental states, ensuring safe and effective interactions in dynamic, real-world environments.
- Designing explainable decision-making frameworks that make autonomous systems’ behavior intuitive and transparent to human users, fostering trust and collaboration in both dyadic and group settings.
In my current postdoc at CMU, I investigate how robots should communicate their decision-making process, abstracted as reward function weights, in human-robot groups. My work focuses on designing machine teaching frameworks that adaptively explain robot behavior to groups by modeling their individual and aggregated group beliefs using particle filters. These closed-loop strategies select demonstrations based on evolving team understanding, improving policy learning. I also study how humans respond to unexpected or faulty system behaviors, particularly in automated vehicles. Using multimodal data including facial expressions, audio, and physiological signals I develop techniques to detect subtle emotional cues such as surprise, confusion, and frustration. This line of work enables real-time identification of system-user expectation mismatches and trust breakdowns, laying the foundation for behavior-aware, trust-sensitive autonomy that can adaptively respond to user needs.
Education
- Ph.D. in Mechanical Engineering at UM, Ann Arbor, in 2021.
- M.S. in Mechanical Engineering at UM, Ann Arbor, in 2018.
- B.E. in Production Engineering at Anna University , India in 2013.
Research Interests
human behavior modeling, explainable decision-making, group human-AI/robot interactions, trust in AI/robots, interaction-aware control, human-AI/robot communication, human-AI/robot teaming, socially intelligent agents
news
| Oct 24, 2024 | Presented at IROS 2024 the work on explainable robot decision-making in groups, titled Understanding Robot Minds: Leveraging Machine Teaching for Transparent Human-Robot Collaboration Across Diverse Groups. |
|---|---|
| Apr 12, 2024 | Gave an invited talk at the Microsoft Leaders in Robotics and AI Seminar, University of Maryland, on machine teaching for transparent decision-making in human-robot teams. |
| Mar 07, 2024 | Presented at the HRI 2024 Workshop on Explainability in Human-Robot Collaboration, sharing work on modeling human learning from robot demonstrations. |
| Oct 25, 2023 | Co-organized the AAAI Fall Symposium on Agent Teaming in Mixed-Motive Situations. |