Path Planning for Mobile Robot by Sequential Generation of Valley and Saddle-shaped Potential Fields
Mahiro Koike, Fumiaki Takemori · 2024
In recent years, autonomous mobile robots that perform cleaning and transportation tasks have been increasingly implemented in society in order to address labor shortages and reduce labor costs.To operate such a robot, it is necessary to obtain information about the surroundings and plan an optimal path according to the environment. In this paper, we discuss the usefulness of a trajectory planning method that applies an artificial potential method that adapts to local environmental information around a robot.The artificial potential method is a control law that allows a robot to move along a gradient by specifying the coordinates and direction of a candidate passage and setting a virtual gradient in that direction.A range sensor is used to estimate the width and direction of the passageway from the robot's perspective.In this study, assume that the map information of the passage is unknown, the coordinates and shape characteristics of the drivable passage is estimated by sensing the real environment with a range sensor, and then a method to generate an appropriate potential field for that local passage is proposed. In experiments, it is shown that seamless robot path planning can be achieved by combining a two types of potential field.