VESC-SLAM: A Visual Information Enhanced Multi-Robot Synchronous Collaborative SLAM

The collaborative Simultaneous Localization and Mapping (SLAM) system between unmanned aerial vehicles and unmanned ground vehicles in low-light environments faces the problem of poor quality such as blurred and missing visual information. This study constructs a collaborative SLAM system based on deep neural networks to solve this problem. The system significantly improves the positioning a ccuracy and map construction quality of unmanned aerial vehicles and unmanned ground vehicles under low-light conditions through the cooperation of the front-end image enhancement module, the collaborative SLAM core processing module, and the intelligent data collaboration and resource optimization module. The system has been validated on the Euroc dataset processed by darkening and its performance has been compared with other robot cluster collaborative SLAM.