Papers
5
Total Citations
27
H-Index
4
About
Shivam Vats is a roboticist focused on enabling robots to autonomously acquire new skills and solve tasks over extended periods—a challenge central to lifelong robotic manipulation. His research bridges task planning, skill learning, and human-robot collaboration, with a particular emphasis on search-based planning methods that leverage learned skill effect models. His most influential work (2022, 14 citations) introduces a framework that allows robots to plan and replan using learned models of skill outcomes, avoiding the rigid assumptions of prior approaches. Vats also addresses recovery from execution failures through model predictive meta-reasoning (2023), enabling robots to efficiently learn and select recovery strategies in uncertain environments. Earlier work on learning to avoid local minima in static environments (2017) improved planning algorithm performance, while his recent research on optimal interactive learning via facility location planning (2025) tackles multi-task collaboration, reducing user burden as robots adapt to novel tasks and preferences. With contributions spanning planning, learning, and human-robot interaction, Vats is advancing the frontier of adaptive, long-deployment robotic systems.
Research Focus
Key Achievements
Top Papers
- 1
- 2Efficient Recovery Learning using Model Predictive Meta-Reasoning4 citations · 2023
- 3Learning to Avoid Local Minima in Planning for Static Environments4 citations · 2017
- 4
- 5Optimal Interactive Learning on the Job via Facility Location Planning1 citations · 2025