屋外環境下で用いる自律移動システムに関する研究：第58報

-自己位置に対する確信度の導入による経路逸脱認識手法の構築-

抄録

We describe recognition of deviating from the route method by monocular camera image. Recently major self-position estimation methods of autonomous mobile system(AMS) use accurate map and observation information (distance, coordinates, etc.). However, these has some problems, high generating map cost, recognition miss matching and deviating from the route etc. We propose the self-position estimation method like human. This method uses monocular camera image. AMS detection distinctive features in surrounding environment from image and estimate self-position. Moreover, when it is different from the assumed route, it returns to the correct spot and route using parameter of correctness of self-position. The result of a simulation shows this method is possible to return correct route.