Learning from Demonstration with State Based Obstacle Avoidance for Mobile Service Robots

摘要

Learning from Demonstration (LfD) is a technique for teaching a system through demonstration. In areas like service robotics the robot should be user friendly in terms of coding, so LfD techniques will be of greater advantage in this domain. In this paper two novel approaches, counter based technique and encoder based technique is proposed for teaching a mobile service robot to navigate from one point to another with a novel state based obstacle avoidance technique. The main aim of the work is to develop an LfD Algorithm which is less complex in terms of hardware and software. Both the proposed methods along with obstacle avoidance have been implemented and tested using Player/Stage robotics simulator.