Designing a Mobile Manipulator and Motion Planning for Autonomous Navigation With A* and Q-Learning Algorithms

Abstract Mobile manipulators have a crucial role in large manufacturing industries, production houses and retailer shops. Although fixed manipulator arms are very familiar in industries and production houses, a manipulator arm mounted on a movable robot equipped with the feature of autonomous navigation is still unfamiliar in this age. In this research, we propose an autonomous navigation method with A* and Q learning algorithm for a 6-DOF manipulator robot arm that is mounted on a movable 2-wheeled robot and the robot can find its path autonomously in a dynamic environment where knowledge of the environment was not known to the robot previously. The robot scans its surroundings with laser scanners and generates a map of its environment in ROS. The robot’s ability to move from its current location to a desired destination while avoiding obstacles and optimizing its trajectory depends heavily on its motion planning. For motion planning of the manipulator robot graph based and reinforcement learning based algorithms can be integrated. Incorporation of these algorithms helps the manipulator robot to find its path to a known or unknown environment. The result demonstrates the environment map generation using laser scans in RViz and applying A* and Q-Learning algorithms the robot finds its path to move from its initial state to its goal state on a generated map.