Targeting and tracking algorithm of humanoid soccer robot

The target location and tracking algorithm was improved and a novel color recognition algorithm was proposed for autonomous robot based on the embedded structure of humanoid soccer robot system. Firstly, the common color identification method was improved by adding the statistical learning process of targeted color features in order to make the threshold of the object color features set by the soccer robot automatically instead of PC. Secondly, the target locating system based on embedded monocular vision system was simplified. The target tracking algorithm based on Camshift was improved by extending the first frame searching area or adding the shape check method. The real-time and autonomous requirement was met.Finally, the simulation and experiment on an 18-degree-freedom humanoid soccer robot was done. In a certain light range, the real-time performance and accuracy of the algorithm were verified. The results show that the improved target localization, tracking algorithm and color identification process have a high performance.