Abstract
Our proposed solution combines point cloud registration with inertial estimation, forming a factor graph that simultaneously optimizes the pose and IMU biases. We developed a zero velocity detection module to ameliorate ICP drift when the agent is static and observes moving objects, as well as a dynamic motion detector to help the system to converge in the presence of aggressive motion. The robustness of our approach has been validated with the datasets from the Hilti SLAM Challenge. Show more
Publication status
unpublishedSubject
RoboticsOrganisational unit
09570 - Hutter, Marco / Hutter, Marco
More
Show all metadata
ETH Bibliography
yes
Altmetrics