Image Processing Based Control System for Wheeled Mobile Robots

Abstract

The paper presents the design and tests of a wheeled mobile robot. The construction of the machine is based on a differential drive. A three-layer structure of the robot control system has been developed. The lowest layer is based on an 8-bit microcontroller and its purpose is to control motors and sensors of the robot. The medium layer is based on a 32-bit ARM Cortex M4 microcontroller. There has been implemented an x86 architecture computational unit as the main part of the highest layer of the control system. A control software developed for this computer is responsible for the acquisition and processing of an image stream from the camera mounted on the robot board. The software has been implemented using Aforge.NET library. Selected image processing filters have been tested in order to enable object tracking and gesture recognition in the mobile robot control system. Finally, a power consumption of the computational unit operating in several developed control modes has been measured.