Dynamic Obstacle Avoidance on Middle Size League Robot Soccer ERSOW Using Subtargets

Abstract

This paper presents the generation of a virtual obstacle to enhance subtargets algorithm that is implemented on robot soccer. One thing that we need to consider is the behavior of our robot when dealing with obstacles. The most challenging ability that must be owned by the robot is to avoid obstacles on the field as fast as possible. Obstacle avoidance skills must be adaptive to highly dynamic movement of the opponent robot and must have cheap computation. Subtargets algorithm is the right answer with considering the speed of each object and the environment of robot soccer. Speed will be converted to several virtual obstacles using a linear constant velocity method and will be input for subtargets algorithm. Based on the result of the experiments have been done. This algorithm worked great if both our robot speed and the virtual obstacles of the opponent robot speed are not exceeding the limits. Robot speed can reach 70 cm/s. The robot has a 100% rate of success to avoid both multiple static obstacles and single static obstacles flawlessly. In the second experiment, we try with different setpoint speeds of virtual obstacle which is 0.25 m/s, 0.6 m/s, 0.65 m/s and the robot can avoid correctly with a 100% rate of success. Third, we increase the virtual obstacles speed until the limits on 1 m/s, the robot is decreasing the rate of success to 90%. This trend will get worst on the rest of the speed if we keep increasing it.