Ravankar Abhijeet, Ravankar Ankit A, Kobayashi Yukinori, Emaru Takanori
Faculty of Engineering, Lab of Robotics and Dynamics, Hokkaido University, Sapporo 060-8628, Japan.
Sensors (Basel). 2017 Jul 5;17(7):1581. doi: 10.3390/s17071581.
Large scale operational areas often require multiple service robots for coverage and task parallelism. In such scenarios, each robot keeps its individual map of the environment and serves specific areas of the map at different times. We propose a knowledge sharing mechanism for multiple robots in which one robot can inform other robots about the changes in map, like path blockage, or new static obstacles, encountered at specific areas of the map. This symbiotic information sharing allows the robots to update remote areas of the map without having to explicitly navigate those areas, and plan efficient paths. A node representation of paths is presented for seamless sharing of blocked path information. The transience of obstacles is modeled to track obstacles which might have been removed. A lazy information update scheme is presented in which only relevant information affecting the current task is updated for efficiency. The advantages of the proposed method for path planning are discussed against traditional method with experimental results in both simulation and real environments.
大规模作业区域通常需要多个服务机器人来实现覆盖和任务并行。在这种场景下，每个机器人都保留自己的环境地图，并在不同时间为地图的特定区域提供服务。我们提出了一种针对多个机器人的知识共享机制，其中一个机器人可以将在地图特定区域遇到的地图变化（如路径堵塞或新的静态障碍物）告知其他机器人。这种共生信息共享使机器人无需明确导航到那些区域就能更新地图的偏远区域，并规划高效路径。提出了一种路径的节点表示法，用于无缝共享堵塞路径信息。对障碍物的短暂性进行建模，以跟踪可能已被移除的障碍物。提出了一种惰性信息更新方案，即仅更新影响当前任务的相关信息以提高效率。针对传统方法，在模拟和真实环境中通过实验结果讨论了所提方法在路径规划方面的优势。
