Vision system for robot´s speed and position control

Machine vision is an important tool used in robotics development to interact with its surrounding and operate autonomously in both known and unknown environments. This paper describes a robot control system for target tracking and following using machine vision. The system allows tracking of another agent equipped with a marker of distinctive colour, simultaneously keeping distance and adjusting its speed accordingly. The leading agent may be a person or another robot, or any other moving object. The system was implemented on a `Pioneer´ 3A-T robot running with Ubuntu, ROS and Matlab on the onboard computer. Matlab is interfaced with ROS to control all the robot´s actuators. The robot follows the target along its path and maintains a specified distance to the target.