Stair Feature Recognition and Modeling based on Binocular Vision

Autonomous obstacle crossing for mobile robots is an important area of research. The premise of autonomous obstacle crossing is the ability of a robot to sense the environment and determine the size of obstacles. This paper uses stairs as the research object, which is the most common and complex obstacle model in structured environments. By analyzing the structural features of stairs, a method to extract the geometric features of stairs from a point cloud is proposed. From the geometric dimensions of the stairs, the data related to robot obstacle crossing are extracted to construct an obstacle model. The experimental results show that this method can effectively extract the characteristics of stairs and provide an obstacle model that can meet the requirements for obstacle crossing.