Papers
7
Total Citations
910
H-Index
6
About
Bart Selman is a leading researcher in artificial intelligence, with a focus on human activity recognition, robotic perception, and unsupervised learning. His pioneering work on unstructured human activity detection from RGBD images, notably his 2012 paper with 528 citations, established low-cost, reliable methods for recognizing complex human actions in home environments using sensors like the Microsoft Kinect. Selman’s contributions extend to developing the Watch-n-Patch system, which learns actions and temporal relations without supervision, and the Watch-Bot, a robotic assistant that detects forgotten tasks and prompts users with a laser pointer—showcasing practical AI for assistive robotics. He has also advanced haptic perception through optoelectronic robotic flesh, enabling embodied AI to sense deformation, temperature, and vibration. His research on automated curriculum strategies for solving hard planning problems, such as Sokoban, demonstrates his impact on deep reinforcement learning. With over 900 citations across his most influential works, Selman’s innovations bridge perception, planning, and human-robot interaction, making him a key figure in developing intelligent, context-aware systems for everyday assistance.
Research Focus
Key Achievements
Top Papers
- 1Unstructured human activity detection from RGBD images528 citations · 2012
- 2Human Activity Detection from RGBD Images273 citations · 2011
- 3
- 4Watch-n-Patch: Unsupervised Learning of Actions and Relations23 citations · 2017
- 5Watch-Bot: Unsupervised learning for reminding humans of forgotten actions14 citations · 2016
- 6
- 7Watch-n-Patch: Unsupervised Learning of Actions and Relations4 citations · 2016