Artificial vision system for differential multiples robots

Multi-robot systems (MRS) have become a promising area in mobile robotics, in applications in which a coordinated action is required. This paper presents an Artificial Vision System (AVS) for a multi-robot platform, developed at the Universidad Nacional de Colombia (at Medellín) with academic and pedagogical aims. The artificial vision system may be viewed as a subsystem of the MRS and is composed by three basic elements: The environment, the camera, and the processing unit. The Artificial Vision System implements color spaces, threshold techniques, and noise elimination for obtaining in real time the relative positions of the robots in the working area. 200 instance images were used as testing set. The obtained algorithm is able to suitably eliminating image noise, and extracting and identifying labels of the robots for around 80% of the testing set. Beyond that, the implemented system may serve as an experimentation platform for subjects as diverse as image processing, control, instrumentation and robotics.