Alexander D. Stewart

Robotics Research (United States), University of Oxford

Papers

Total Citations

149

H-Index

About

Alexander D. Stewart is a leading researcher in robotic vision and autonomous navigation, with a core focus on long-term visual localisation for mobile robots. His most influential work, "Shady dealings: Robust, long-term visual localisation using illumination invariance" (2014, 141 citations), tackles a critical challenge in outdoor robotics: maintaining reliable localisation across changing lighting conditions and extended periods. By developing methods to discover robust feature correspondences between recorded and live stereo images, Stewart’s approach enables robots to operate persistently in real-world environments where shadows, weather, and time of day would otherwise cause failure. This contribution has become a foundational reference for researchers working on visual place recognition and lifelong mapping. Stewart also explored low-cost localisation alternatives in his thesis, "Localisation using the appearance of prior structure" (2014, 8 citations), where he demonstrated that monocular cameras could achieve accurate pose estimation by leveraging a coloured 3D pointcloud prior—a promising direction for affordable autonomous systems. His work bridges the gap between theoretical robustness and practical deployment, making him a key figure in advancing field-ready robotic navigation.