Research And Implementation On Environmental Perception And Obstacle Avoidance Strategy Of Humanoid Soccer Robot

Robot soccer provides a standard platform for researchers to study the related technology which is a landmark and challenging academic subject, together with interesting and participation, obtains the global scholars and robot lovers’attention.So robot soccer is the study hotspot in recent years.In the dynamic game environment,the result of the competition will be affected because of the environment perception and the collision caused by robots,therefor studying the environment perception and obstacle avoidance strategy has important significance and application value.In this thesis, a number of the related literatures are consulted and relevant technology are studied.The robot environmental perception and obstacle avoidance strategies are studied.In robot environmental perception aspect,first, the target’s seed is selected to segment the target (ball and goal) by using the region growth algorithm.The effectiveness and accuracy of the algorithm is analysed.Then,in the Nao robot vision system, the distance-computed model is constructed and the object repair algorithm is realized in order to solve the distance’s error caused by the object shelter.The accurate distance between the ball and the robot can be got. The robot position is adjusted according to the angle deviation between the robot and the goal’s gravity center in order to get the best shooting angle, which is important to gain score. Third, for football tracking problem, the algorithm based on the environmental perception and the UKF algorithm are studied when robot is in different situation such as robot is static or moving. The accuracy and the error of the tracking are analyzed.Some experiments are done to verify the effectiveness.For testing the tracking result of the UKF when robot and the ball are all moving, some simulation experiments are done to verify the UKF’s effectiveness.In robot obstacle avoidance strategy aspect,some obstacle avoidance strategy is designed based on multi-sensor information,according to the fuzzy logic rules,obstacle avoidance behavior is autonomous realized.The effectiveness of the method is verify by the experiments.Experimental results show that the target segmentation algorithm can effectively achieve the goal’s division, the distance between the robot and the target can be got by constructing the monocular vision measurement model which enhances the robot’s environmental perception ability.By combining the sensor information, obstacle avoidance strategy can effectively reduce the collision between the robots.