Image-Based Humanoid Robot Pose Recognition System

In this paper, an image-based humanoid robot pose recognition system is designed and implemented to recognize humanoid robot pose and some objects position in the field. Robot Operating System (ROS) is used to establish the software development framework for the image recognition system. The image recognition system includes feature matching, region of interest (ROI) capture, background subtraction, edge detection, and object segmentation. Then the Inverse Perspective Mapping (IPM) is used to determine the location of humanoid robot and some objects. Experimental results show that the proposed robot pose recognition system can simultaneously recognize and track both of obstacle, humanoid robot and its pose.