Mobile Robot Motion Planning through Obstacle State Classifier

Abstract

Motion planning for obstacle avoidance is one of the essential capabilities for autonomous navigation by mobile robots. We have thus far proposed a motion planner based on CNN with an LSTM block through mediated perception. As a result, a mobile robot based on the motion planner was enabled to avoid a walking person as a dynamic obstacle. In this paper, the robot further plans different avoidance motions depending on the velocity of the dynamic obstacle. For this challenge, an obstacle state classifier based on CNN is used ahead of the motion planner. A depth-difference image generated from two depth images is used as the input to the classifier. A classified state that indicates the velocity of the obstacle is fed as the input to the following motion planner. Finally, navigation experiments show that the robot based on the motion planner with the obstacle state input is able to plan different avoidance motions for a person walking slowly or fast through the obstacle state classifier.