Small Robot Ladder Environment, Visual Navigation Research

Intelligent robot is a hotspot research field all over the world, and vision navigation is one of key techniques. To meet the demand of vision navigation for miniature mobile land robot, the vision navigation under stair environment is researched in this paper.The main task of paper is to extract step edges effectively and reliably from stair image. Therefore, following processing steps are adopted: stair image preprocessing, stair edges detection, step edges extraction. For first step, close operator of morphology is applied so as to smooth image and remove dark details of stairs, which can get more continuous step edges. In the second step, canny operator is used to detect stair edges, which can preserve more step edges and supply enough information for subsequent process. In the last step, a new approach of line detection is presented based on line mask to eliminate many non-line edges. Finally, an effective approach is designed to get the direction of stair steps and locate step edges based on prior knowledge of stairs.The experimental results show the approach presented in this paper can be applied to many different types of stairs and is not sensitive to lighting condition. It is effective for both stairs indoors and outdoors. Besides, it can also adapt conditions of part abrasion, water mark and shadow. In a word, it is a robust, reliable and real-time approach that can meet the requirement of vision navigation of miniature mobile land robot.