Papers
14
Total Citations
330
H-Index
7
About
Chaitanya Mitash is a robotics and computer vision researcher whose work centers on 6D pose estimation, object tracking, and perception systems for robotic manipulation. His research addresses fundamental challenges in enabling robots to understand and interact with the physical world, particularly in cluttered, real-world environments where occlusions and limited labeled data pose significant obstacles. Among his most influential contributions is SE(3)-TrackNet, a data-driven approach for 6D pose tracking that bridges the gap between synthetic and real domains, garnering over 123 citations. His early work on self-supervised object detection using physics simulation and multi-view pose estimation — cited over 100 times — demonstrated how robots could learn to recognize objects without extensive manual annotation, a critical advance for scalable robotics applications. His physics-based reasoning for pose estimation in clutter further refined multi-object understanding under occlusion. More recently, Mitash contributed to ARMBench, a large-scale benchmark dataset for warehouse robotic manipulation developed in collaboration with Amazon, reflecting both his academic rigor and real-world industry impact. His research on online object model reconstruction highlights a commitment to lifelong learning systems that improve through experience. Across his portfolio, Mitash has consistently advanced the frontier of robust, practical robot perception.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3ARMBench: An Object-centric Benchmark Dataset for Robotic Manipulation32 citations · 2023
- 4Physics-based scene-level reasoning for object pose estimation in clutter20 citations · 2019
- 5
- 6
- 7That and There: Judging the Intent of Pointing Actions with Robotic Arms9 citations · 2020
- 8
- 9
- 10