Accurate Localization Method in Multirobot Cooperative SLAM Systems under Resource Constraint

The enhancement of simultaneous localization and mapping (SLAM) is crucial in the evolving field of robotics, which often leads to poor trajectory localization robustness, especially under resource constraints. This study presents a relocalization method based on a historical data model and a local optimal frame-dropping strategy with a communication buffer aimed at improving the reliability and accuracy of SLAM in such challenging environments. To maintain the continuity of the back-end trajectory, the historical keyframe data were used to recalculate poses when tracking is lost. Additionally, to address data loss issues due to network unreliability, the concept of similarity measurement between frames is introduced to model the buffer problem as a local optimization issue. In real-world remote scenarios, using the Technical University of Munich public dataset as input for the SLAM’s front end demonstrates substantial improvements over the existing Centralized Collaborative Monocular SLAM algorithm, where the success rate of system tracking has increased by 4 times, and the trajectory accuracy has increased by 90.27%. These results highlight the proposed method’s effectiveness in enhancing the robustness and accuracy of trajectory localization in resource-constrained settings, marking a substantial advancement in the field of visual SLAM.