Max Robotics — Global Service Robot Directory + US Certification

About

Martin Brossard is a robotics and autonomous systems researcher whose work sits at the intersection of state estimation, sensor fusion, and deep learning for robot navigation. His research focuses on inertial navigation, Kalman filtering on Lie groups, and learning-based approaches to improve the reliability of odometry and localization systems in GPS-denied environments. Brossard has made significant contributions to the mathematical foundations of filtering on non-Euclidean spaces, co-developing Unscented Kalman Filtering on Lie Groups (94 citations) and extending these methods to visual-inertial SLAM (74 citations) and consistency-aware EKF design (70 citations). His work on exploiting geometric symmetries to address filter inconsistency has particular relevance for real-world SLAM deployments. Perhaps his most impactful contribution is applying deep learning to IMU denoising, demonstrating that neural networks can substantially improve open-loop attitude estimation — a result that garnered 148 citations and helped establish learning-based inertial odometry as a mainstream research direction. His complementary work on wheel odometry (100 citations) and ICP covariance estimation (65 citations) further underscores his broad influence on robust mobile robot localization. Collectively, his publications represent a cohesive effort to make autonomous navigation more accurate, principled, and deployable under challenging real-world conditions.