About Me
I am a PhD candidate in Robotics at the University of Michigan - Ann Arbor, working in the Fluent Robotics Lab with Prof. Christoforos Mavrogiannis. I’ve always been fascinated by how easily humans move around one another without really thinking about it, and I want robots to be able to do the same. My research is in motion planning, control, and human-robot interaction, mostly focused on social navigation: getting robots to move through spaces full of people in ways that feel predictable and comfortable, rather than mechanical.
My current work builds MPPI-based controllers for legible robot motion. In Rethinking Legibility in Social Robot Navigation, we found that signaling a robot’s immediate choice, like which side to pass on, works better than signaling its final goal, and holds up even when people are distracted.
More broadly, my work follows one thread: a robot moving among people acts under actionable uncertainty. That uncertainty comes from context, like whether a person is distracted or attentive, or whether a blind corner hides an oncoming pedestrian. I am interested in characterizing that uncertainty and in using the robot’s own motion to act on it: how a robot moves decides both what it gets to observe and what people infer about it.
A related line of my work is on human motion prediction and the data behind it. This includes studying how prediction quality shapes navigation performance, and building the datasets (ACME, Bi3) and scenario-based testing tools (SocRATES) these models depend on.
Before Michigan, I was a Research Assistant at the CLeAR Lab, National University of Singapore, with Prof. Harold Soh, working on diffusion-based planning and social navigation for legged robots. During my undergrad at BITS Pilani, K.K. Birla Goa Campus, I spent a year at CMU’s Biorobotics Lab with Prof. Howie Choset and Prof. Robin Murphy, working on legged-wheeled robot control and 3D perception for search and rescue.
More on my Publications, Experience, and CV pages, or reach me at prgoyal@umich.edu.
News
- HRI 2026 (LBR): Rethinking Legibility in Social Robot Navigation
- HRI 2026: How Human Motion Prediction Quality Shapes Social Robot Navigation Performance
- IEEE RA-L: SocRATES: Automated Scenario-based Testing of Social Navigation Algorithms
- ICRA 2026: Bi3: A Biplatform, Bicultural, Biperson Dataset for Social Robot Navigation
- RSS 2024 workshop: Runner-up Best Paper Award for our scenario-testing paper