Abstract
A new path planning method for mobile robots is proposed. When a mobile robot navigates, an error in estimated position by dead-reckoning of the robot increases with the traveling distance. Generally, environment recognizing functions of a robot are limited, and hence the positioning error gives a risk of collision with objects in the envi-ronment. To reduce this error, the robot should detect the natural or artificial landmark in the environment by its own sensors, and adjust it's estimated position. It is necessary for a mobile robot navigation to adjust it's position in suitable intervals. However, there are not so many landmarks to adjust robot's position in real environment. Hence planning the path with good landmarks is necessary for mobile robot navigation to a goal. On the other hand, it is nonsense to take huge cost to detect such landmarks. In this case, trade-off between sensing costs and good landmarks is important.
In this paper, the cost function for the mobile robot's path which minimizes the risk of collision related with sensing cost, and the algorithm to find the optimal path and sensing points are proposed. The authors implemented the proposed algorithm on real robot system, and made several experiments to verify this algorithm.
