Papers
8
Total Citations
457
H-Index
6
About
Brian P. Williams is a versatile robotics and artificial intelligence researcher whose work spans computer vision, autonomous planning, and robotic manipulation. He is perhaps best known for his foundational contributions to monocular Simultaneous Localisation and Mapping (SLAM), particularly his pioneering work on real-time relocalization — the ability of camera-based systems to recover their position after tracking failure. His 2007 paper on real-time SLAM relocalization has garnered 224 citations, establishing him as a key figure in making vision-based pose estimation robust enough for real-world robotics and augmented reality applications. His 2011 follow-up extended these ideas further, addressing map reuse and loop closure challenges that had long limited practical deployment. Beyond visual SLAM, Williams has made notable contributions to risk-bounded motion planning under uncertainty, mixed discrete-continuous activity planning through convex optimization, and learning-based manipulation. His early work even touched on space robotics, contributing to the design of a softball-sized autonomous flying robot for microgravity spacecraft environments. More recently, he has explored reinforcement learning approaches to sparse-reward manipulation tasks. Across diverse research threads, Williams consistently bridges theoretical rigor with practical robotics challenges, making his body of work valuable reading for students working at the intersection of perception, planning, and autonomous systems.
Research Focus
Key Achievements
Top Papers
- 1Real-Time SLAM Relocalisation224 citations · 2007
- 2Automatic Relocalization and Loop Closing for Real-Time Monocular SLAM97 citations · 2011
- 3
- 4ScottyActivity: Mixed Discrete-Continuous Planning with Convex Optimization36 citations · 2018
- 5
- 6R2D2 in a softball21 citations · 2000
- 7Simultaneous Localisation and Mapping Using a Single Camera3 citations · 2009
- 8