Sumit Patidar
Papers
2
Total Citations
43
H-Index
2
About
Sumit Patidar is a roboticist advancing the frontier of dexterous, multi-task manipulation. His primary research areas span hierarchical task planning, diffusion-based policy learning, and in-hand object reconfiguration. Patidar’s most impactful contribution is the **Hierarchical Diffusion Policy (HDP)** , a novel framework that factorizes robotic manipulation into two levels: a high-level agent predicting a distant next-best end-effector pose, and a low-level goal-conditioned controller. This design enables robust, kinematics-aware performance across diverse tasks, earning **39 citations** since its 2024 publication and positioning HDP as a key reference in imitation learning for robotics. In parallel, his work on **in-hand cube reconfiguration** (2023) demonstrates a stripped-down, robust approach that challenges assumptions about the complexity of dexterous manipulation. By simplifying planning, control, and perception, Patidar reveals fundamental insights into the problem’s inherent difficulty. His research is distinguished by its practical elegance—achieving general, reliable performance without unnecessary complexity. For students and researchers, Patidar’s work offers a clear lesson: the most profound advances in robotics often come from asking what can be removed, not added.
Research Focus
Key Achievements
Top Papers
- 1
- 2In-Hand Cube Reconfiguration: Simplified4 citations · 2023