Visual Tracking Control of Humanoid Robot

This paper realizes the independent complex route tracking based on the NAO robot hardware platform and an image processing technology. The camera captures an image and extracts a path through a series of image processes, such as threshold, filtering, and edge detection. The path recognition algorithms and its edge array were used to calculate these relative parameters that drive the anthropomorphic robot for complex route walking for autonomous tracking. These paths include the straight-route curves and the cross routes, which were identified by the different features accurately. Particularly, the slope-matching method is proposed to enable it to track in accordance with the established rules when meeting the cross routes. These methods separately calculate the slope from the intersection to paths in all directions, and then, the matching rate can be obtained by a predefined matching formula in terms of choosing the best path forward. All complex route tracking experiments were conducted by an anthropomorphic robot. The experimental results show that the robot imposed a strong anti-interference ability to filter out the noise and have accurate complex route tracking, which is significant for the anthropomorphic robot visual guide.