A SVM-PSO Classifier for Robot Motion in Environment with Obstacles

Abstract

Support vector machines (SVM) is a popular method of classification problems. In this study, the sensor data obtained from the movement of the robot SCITOS G5 along the road is classified. Sensor data includes numerical values of the movements of the robot against obstacles. C and Gamma values were optimized with Particle Swarm Optimization (PSO) from the most important parameters affecting the accuracy of classification in SVMs and the most appropriate values were used. In this paper, it has been investigated whether any robot can exhibit humanoid movements with SVM method against an obstacle.