Bharath Sankaran

University of Southern California, University of Pennsylvania

Papers

Total Citations

H-Index

About

Bharath Sankaran is a robotics and computer vision researcher whose work sits at the intersection of autonomous perception and intelligent decision-making. His research focuses primarily on active perception, view planning, and robot autonomy — areas that address fundamental challenges in enabling machines to understand and interact with complex environments. Sankaran's most influential contribution is his work on nonmyopic view planning for active object classification and pose estimation, which has garnered 82 citations and represents a significant advance in the field. Rather than relying on single-image processing — an approach inherently limited by occlusions and visual ambiguity — his research develops principled strategies for robots to actively plan sequences of viewpoints that maximize the quality of object detection and pose estimation. This forward-looking, multi-step planning approach moves meaningfully beyond reactive, myopic methods that dominated prior work. Beyond perception, Sankaran has contributed to the challenge of long-term robot autonomy through his research on failure recovery with shared autonomy, addressing the practical need for robots to perform tasks with reliability and robustness in personal robotics contexts. Together, his body of work reflects a coherent vision: building robots that perceive their environments intelligently and recover gracefully when things go wrong — essential capabilities for real-world autonomous systems.