In context
In 2024, brain-inspired navigation models such as RatSLAM were gaining attention for large-scale mapping due to their low computational and storage requirements. However, single-robot systems still faced efficiency bottlenecks in expansive environments, prompting research into multi-robot collaborative approaches to accelerate map construction.
What was reported
A study published in the Chinese journal Robot presented a multi-robot cooperative RatSLAM system designed to improve mapping efficiency in large environments. The system uses a centralized communication architecture based on ROS, allowing robots to share visual and pose information in real time. An overlapping region detection method was developed to identify when robots have visited the same location, enabling data association and relative pose estimation between experience nodes.
The authors introduced an improved graph relaxation algorithm that fuses local maps from multiple robots into a globally unified cognitive map. The method was validated on public datasets and in real-world physical experiments. Results showed that compared to a single robot, the multi-robot system increased average mapping efficiency by 45% while maintaining higher map accuracy and requiring less storage. The proposed approach also outperformed existing map fusion methods.
The system builds on Fast-iBoW-RatSLAM, using ORB features and an incremental binary search tree for fast image retrieval. A dynamic threshold and continuous sequence detection enhance the reliability of overlap detection, making the system adaptable to varied environments.
Why it mattered
This work demonstrated that brain-inspired SLAM can be effectively extended to multi-robot scenarios, offering a practical path to faster, scalable mapping for industrial automation. The reduced computational load and improved efficiency make it suitable for long-duration tasks in large facilities, where collaborative robots can cover ground more quickly without sacrificing map quality.
“Compared to single robot, the average mapping efficiency of multiple robots is increased by 45%, meanwhile maintaining higher map accuracy and requiring less storage.”
Source: 《机器人》期刊 (robot.sia.cn) · Published 2024-07-20 · “多机协同的类脑同步定位建图方法”