Object search, recognition and following in a cluttered environment

Abstract

The popularity of mobile robots is increasing as we are becoming more aware of how powerful and useful these machines can be in everyday life. Autonomous robots are even more attractive as these can complete jobs assigned to them without the need of any supervision. Such robots should be capable of operating in around cluttered environments without damaging themselves or creating a hazard to humans sharing the same space. This project endeavors to create a system where the mobile robot first searches for a desired object using a vision system, and when the object is found, the robot starts to approach it and at the same time avoid any obstacles lying in its path towards the target. As the robot arrives at a pre-defined distance away from the target object, it stops and if the target moves, the robot is able to continue following it. This whole system consists of a number of algorithms that work harmoniously together to perform multiple tasks depending on the need at that particular circumstance. A thorough study about existing algorithms has been carried out at the start of the project. Some of these algorithms were simulated and studied into more detail. For the ‘search’ task, no existing algorithm which satisfies the requirements could be found and therefore a new algorithm was developed to accommodate all the requirements mainly imposed on the system by the hardware of the robot used. The designed system was successfully implemented on the Khepera III mobile robot.