Teach and Repeat for Path Planning Using Bézier Curves

In the service robots field, a mobile robot must be able to navigate and repeat previously defined paths. This paper aims to develop a technique capable of teaching and repeating (T &R) a path taken by a mobile robot. We evaluated using a differential robot. The developed system is integrated with the ROS framework and uses the mobile base, which is built with self-balancing scooter components. When changes need to be made in the environment, a user must generate a new path by teleoperating the robot, and the system can repeat new paths autonomously. The proposed approach consists of using Bezier curves to represent the paths taken by the user. The proposed method implements a T &R technique to navigate a mobile robot. Results show the worst maximum average of percentage error (MPE) using this method was only 1.361 % in adapting the Bezier curve to the points where the robot was taught, satisfactorily representing the path. In the repeating process, the worst MPE was 9.983%.