障害物が疎らな未知環境で位置・姿勢・センサ誤差を克服するセンサベーストナビゲーション

抄録

In the last decade, many sensor-based navigation algorithms have been proposed, which keep convergence of a mobile robot to its destination in a 2-D uncertain environment. From the convergence viewpoint, all the previous algorithms are categorized into three types, i.e., metric, topologic, and geometric algorithms. A mobile robot supervised by each algorithm frequently joins a limit cycle (deadlock) if the robot generates position, orientation, and/or sensor errors. Especially, geometric and topologic algorithms are theoretically damaged by position and sensor errors, respectively, even though they are small enough in a 2-D unknown environment. Therefore, we are obliged to rely upon the metric algorithm if all errors exist. In case that a mobile robot is ready for position, orientation, and sensor errors, we propose a stable metric sensor-based navigation algorithm for overcoming all the errors. The stable algorithm always leads a robot near its destination in a sparse environment whose minimum distance of obstacles is sufficiently larger than the sum of a radius of a mobile robot and position ingredients of the errors. Finally, we ensure convergence of a robot in destination's neighborhood in the proposed algorithm by theoretical proof and experimental result.