Contour detection and localization of intelligent wheelchair for parking into and docking with U-shape bed

Towards the intelligent wheelchair´s parking into and docking with a U-shape bed to expand as a complete size-wide bed, a U-shape contour detection and wheelchair localization method is proposed in this paper. Lines contained in the back-view image of the wheelchair are projected into the bed´s superface plane and furthermore separated into several groups based on conditions of parallelism and perpendicularity. The U-shape contour is searched among the projected lines in each group by model matching, and the matched lines with highest priority and smallest error are determined as the ultimate U-shape boundary lines. Based on these boundary lines´ direction vector and crossing points, the wheelchair´s orientation and position relative to the bed are determined according to relations between the bed coordinate system and the camera coordinate system. By comparing with the MR code method, experimental results validate the feasibility and applicability of the detection and localization method proposed in this paper.