A new method to follow a path on indoor environments applied for mobile robotics

摘要

A new method to follow a path conformed by straight lines, rotation angles and landmarks is presented. This method implements odometry and computer vision in order to achieve desired position. The method can be applied for indoors such as industries, laboratories, hospitals and other closed places. The approach is divided in two stages: data acquisition and autonomous navigation. At first stage, user guides mobile robot through desired trajectory in order to acquire angles, landmarks and distances. Also, an algorithm to acquire landmarks is established. The algorithm implements image descriptors and it is based on Harris corner [1] and Shi-Tomasi [2] algorithms. In second stage, mobile robot replicates path applying odometry for straight lines and SURF algorithm for image feature detector [3]. Two movements are required with the purpose to perform specific task: go forward and rotation so two controllers are designed. Finally, experiments and results are exposed.