Improvement of the Processing Speed of The Robot’s Vision System Using Robot Operating System

摘要

This Research presents the improvement of the processing speed of the vision system at Robot ERSOW using Robot Operating System. In Mobile robot Soccer ERSOW to perform the ability to find the ball, kick the ball, avoid the obstacles, and shooting the ball into the goal, accurate supporting data is required. There are 3 objects that have been detected by this vision system, that is ball detection, robot detection, and obstacles detection. Based on the number of objects detected, vision systems in ERSOW robot require large computations and processing times. Therefore, by using Robot Operating System (ROS), the speed of the object detection process in the ERSOW robot vision system can increase rapidly. In addition, using ROS and the upgraded detection concept will consume less memory computing. In this research, the result of ball detection average FPS will increase by ±17.19% and the average computation of the ball detection will decrease by ±16.67% with the average speed processing ± 18ms. For obstacle detection using ROS, the average FPS will increase by ±32.92% and the average computation will decrease by ±16.10% with the average speed processing ± 13ms. And for the detection of robots using ROS, the average FPS will increase by ±19.07% and the average computation will decrease by ±13.78% with the average speed processing ± 26ms.