Can robots help each other to plan optimal paths in dynamic maps?

Mobile robots need to plan an optimal and collision-free path in complex environments to the various service locations in the map. Robot perception is limited to sensor range and hence robots are only aware of the obstacles within the range of its sensor. Moreover, the knowledge of new obstacles observed by each robot is kept local to itself. Due to these limitations, robots are often unaware of the new obstacles or path blockages in remote sections of the map. We proposed a knowledge sharing architecture in which the robots share the knowledge of new obstacles and blocked paths with each other. This enables the robots to update their map with the new and remote obstacles, and plan efficient paths with the timely information. This eliminates the need for re-planning of paths in traditional navigation methods. Experimental results show that the proposed method is efficient for multi-robot navigation in large and dynamic maps.