Overhead vision system for mobile robot orientation detection

Abstract

Robot cooperation and coordination is absolutely necessary in many industrial applications. The computation of a mobile robot position and orientation is a common task in the area of computing vision and image processing. For a successful application, it is important that the position and orientation of the mobile robot are properly determined. Computing the orientation is not a straightforward technique. Number of methods has already been studied by many researchers. These methods include the concepts of geometric moments, complex moments, and principal component analysis. In this work, a simple procedure for determining the orientation of the mobile robot using overhead vision system is presented and analysed. Cameras are used to capture the images of mobile robot at various orientations. The images are preprocessed and important features are extracted to be used in the proposed methods. In this research, simple methods to extract the features from the preprocessed images are developed. The extracted features are then used as the inputs to a simple feed forward neural network. The orientation of each image is measured manually and used as a target vector. A simple neural network model is developed to estimate the orientation of the mobile robot. Simulation results show that the proposed algorithms can be used to estimate the orientation of the mobile robot accurately.