Multi-Robot Ground Texture SLAM with Communication-Efficient Data Sharing

Kyle M. Hart, Yewei Huang, Brendan Englot

Keywords: Visual SLAM, Multi-Robot

TL;DR: This work presents a method for multi-robot decentralized SLAM that uses ground texture images as the only source of information and accounts for bandwidth constraints when sharing information.

Abstract: When operating in challenging sparse environments such as a ship's deck or an empty road, traditional visual simultaneous localization and mapping (SLAM) algorithms may struggle with the lack of outward facing salient features. We propose a distributed multi-robot SLAM system that exploits the ground texture beneath each robot as a source of features for navigation, and selectively shares this information with other robots to improve SLAM results without overwhelming communication channels. We show with initial experiments that this approach yields more accurate pose estimation than relying on single-robot ground texture SLAM for each robot.

Email Sharing: We authorize the sharing of all author emails with Program Chairs.

Data Release: We authorize the release of our submission and author names to the public in the event of acceptance.

Paper Acceptance: No

Submission Number: 24