A laser-based mobile robot navigation in structured environment

This paper describes the navigation system of the mobile robot ROMANE. A navigation architecture is presented to integrate the tasks which together allow the robot to move in a known or unknown environment with dynamic obstacles. The tasks are distributed to several CPU, under a real time operating system, linked in networks. In order to reach a final goal the robot passes several different sub-goals. A sub-goal is a passage point which must be seen by an external sensor. A laser scanner is used for environment perception. The local map is generated by segment and corner primitives. The subgoal is updated by interpretation of the local map. A path generator then plans a path from robot to sub-goal. To find the goal and to execute the path the robot needs to know its position, therefore two localization tasks (relative and absolute) are developed. The fundamental problem of integration is the computation time of each task. The algorithms are optimized in time to overcome the problem of their integration in the navigation system. The present system can update the sub-goal 4 times per second. This system is developed in order to be transferred and integrated onto an autonomous power wheelchair.