Adaptive Color Mapping for NAO Robot UsingNeural Network

While playing soccer, the main task of the robotvision system is identifying and tracking objects suchas ball, goals, teammate robots and opponent robots.The basis of many object identification methods,particularly those in soccer robots and RoboCupenvironment, is using algorithms based on pixel colorproperties. One of the major problems of these robotsin RoboCup environment is changes in lightingconditions of match environment and in turndifficulty in identification of environment colors andobjects in order to segment the image and identify theobjects. In this paper, a pixel color-base identificationmethod has been suggested using a neural networkfor recognizing the pixels related to each object. Theneural network used in this study has 6 outputneurons for identifying 6 classes of signs includingball, goal, field, field lines, teammate robot andopponent robot. This proposed method has beentested on over 1000 frames of images received byrobot’s camera and different data sets and hasrevealed an appropriate performance withidentification rate of over 90% and error rate of 4%.

Keywords:

humanoid robot , Standard PlatformLeague , vision system , neural network