Papers
16
Total Citations
273
H-Index
8
About
Guy Rosman is a leading researcher at the intersection of robotics, computer vision, and machine learning, with a focus on enabling intelligent systems to perceive, predict, and act in complex environments. His work spans scene understanding, surgical robotics, and autonomous vehicle safety. Rosman made foundational contributions to 3D scene representation with his highly cited work "A Mixture of Manhattan Frames" (65 citations), which relaxed the restrictive "Manhattan World" assumption for man-made environments. In surgical AI, his papers on real-time video segmentation (65 citations) and the SUPR-GAN framework for surgical phase prediction (29 citations) have advanced automated workflow analysis in laparoscopic and robot-assisted surgery. Rosman also pioneered coreset methods for efficient visual summarization in robotics, and developed ShadowCam, a vision-based system for detecting moving obstacles around corners for autonomous vehicles. His work on interpretable trajectory forecasting (MATS) and counterfactual simulation for planner failure discovery addresses critical challenges in human-robot interaction and autonomous vehicle safety. With over 250 citations across these key contributions, Rosman's research consistently bridges theoretical innovation with practical, safety-critical applications in robotics and autonomous systems.
Research Focus
Key Achievements
Top Papers
- 1A Mixture of Manhattan Frames: Beyond the Manhattan World65 citations · 2014
- 2
- 3
- 4Coresets for visual summarization with applications to loop closure24 citations · 2015
- 5
- 6
- 7Information-Driven Adaptive Structured-Light Scanners10 citations · 2016
- 8Variational End-to-End Navigation and Localization9 citations · 2019
- 9Task-Specific Sensor Planning for Robotic Assembly Tasks8 citations · 2018
- 10