Michael Warren
Papers
7
Total Citations
202
H-Index
6
About
Michael Warren is a leading roboticist whose research centers on long-term visual navigation, autonomous localization, and field robotics. His most influential contribution is developing "visual teach and repeat" (VT&R) algorithms that enable robots to autonomously follow manually taught paths over long distances using only low-cost vision sensors. His seminal 2016 paper, "Bridging the appearance gap" (82 citations), directly tackles the critical challenge of environmental appearance change—caused by lighting, weather, and seasons—that has historically limited outdoor robot deployment. Warren also pioneered low-cost stereo vision systems for pose estimation (42 citations) and online stereo rig calibration (41 citations), essential for robust SLAM in long-term autonomy. He co-created OpenFABMAP (14 citations), an open-source toolbox that democratized appearance-based loop closure detection for the robotics community. His work on gimbal-stabilized VT&R (9 citations) further improved localization in rough, unstructured terrain, advancing field-deployable robots for search-and-rescue, agriculture, and border patrol. By systematically addressing the "appearance gap," Warren has helped bridge the divide between laboratory demonstrations and real-world, all-weather autonomous navigation.
Research Focus
Key Achievements
Top Papers
- 1
- 2Unaided stereo vision based pose estimation42 citations · 2010
- 3Online calibration of stereo rigs for long-term autonomy41 citations · 2013
- 4
- 5Graphcut-based interactive segmentation using colour and depth cues10 citations · 2010
- 6
- 7Experimental Comparison of Odometry Approaches4 citations · 2013