An efficient object recognition and self-localization system for humanoid soccer robot

Abstract

In the RoboCup soccer humanoid league competition, the vision system is used to collect various environment information as the terminal data to finish the functions of object recognition, coordinate establishment, robot localization, robot tactic, barrier avoiding, etc. Thus, a real-time object recognition and high accurate self-localization system of the soccer robot becomes the key technology to improve the performance. In this work we proposed an efficient object recognition and self-localization system for the RoboCup soccer humanoid league rules of the 2009 competition. We proposed two methods : 1) In the object recognition part, the real-time vision-based method is based on the adaptive resolution method (ARM). It can select the most proper resolution for different situations in the competition. ARM can reduce the noises interference and make the object recognition system more robust as well. 2) In the self-localization part, we proposed a new approach, adaptive vision-based self-localization system (AVBSLS), which uses the trigonometric function to find the coarse location of the robot and further adopts the measuring artificial neural network technique to adjust the humanoid robot position adaptively. The experimental results indicate that the proposed system is not easily affected by the light illumination. The object recognition accuracy rate is more than 93% on average and the average frame rate can reach 32 fps (frame per second). It does not only maintain the higher recognition accuracy rate for the high resolution frames, but also increase the average frame rate for about 11 fps compared to the conventional high resolution approach and the average accuracy ratio of the localization is 92.3%.