Papers
9
Total Citations
148
H-Index
6
About
Sudeep Pillai is a robotics and computer vision researcher whose work spans 3D perception, depth estimation, visual odometry, and robot manipulation. His research is particularly focused on enabling robots to understand and navigate complex environments through intelligent visual processing, often with minimal supervision. Pillai's most influential contribution, "3D Packing for Self-Supervised Monocular Depth Estimation" (2020, 35 citations), introduced PackNet, a groundbreaking deep network that allows robots to estimate dense depth from a single camera using only unlabeled monocular video — a significant step toward reducing reliance on expensive LiDAR sensors. Complementing this, his work on semi-supervised monocular depth estimation (2019, 24 citations) further advanced self-supervised learning paradigms in computer vision. His early research on learning articulated motions from visual demonstration (2014, 40 citations) demonstrated how robots could interpret and interact with everyday jointed objects like cabinet doors and drawers, earning it his highest citation count. Across his portfolio, Pillai consistently bridges theoretical elegance with practical robotics applications, addressing challenges in stereo reconstruction, SLAM, place recognition, and ego-motion learning. With over 140 cumulative citations, his work has meaningfully shaped how autonomous systems perceive and reason about the physical world.
Research Focus
Key Achievements
Top Papers
- 1Learning Articulated Motions From Visual Demonstration40 citations · 2014
- 23D Packing for Self-Supervised Monocular Depth Estimation35 citations · 2020
- 3High-performance and tunable stereo reconstruction27 citations · 2016
- 4
- 5SLAMinDB: Centralized graph databases for mobile robotics8 citations · 2017
- 6Self-Supervised Visual Place Recognition Learning in Mobile Robots6 citations · 2019
- 7Towards visual ego-motion learning in robots3 citations · 2017
- 8Learning Articulated Motions From Visual Demonstration3 citations · 2015
- 9High-Performance and Tunable Stereo Reconstruction2 citations · 2015