Harshal Shende
Papers
1
Total Citations
2
H-Index
1
About
Harshal Shende is a researcher advancing the frontier of embodied intelligence through robust perception for articulated objects. His work centers on category-level articulation pose estimation, a critical challenge for enabling robots to manipulate, grasp, and interact with everyday objects that move—like drawers, cabinets, and doors. Shende’s key contribution lies in integrating differentiable rendering into pose estimation pipelines, allowing models to reason about kinematic constraints and self-occlusion, which have long plagued articulated object perception. His most-cited paper, “Towards Robust Category-level Articulation Pose Estimation via Integrated Differentiable Rendering” (2025), has already garnered 2 citations, signaling early impact in a rapidly evolving field. By bridging computer vision and robotics, Shende’s research directly addresses the gap between static object recognition and dynamic, real-world interaction. His work is foundational for next-generation autonomous systems that must understand not just what an object is, but how it moves. For students and researchers in robotics, Shende’s approach offers a compelling blueprint for making perception more physically grounded and task-ready.
Research Focus
Key Achievements
Top Papers
- 1