Automated doorway detection for assistive shared-control wheelchairs

This work presents an algorithm for rapid, automated detection of open doorways using 3D point cloud data. The algorithm has been developed in the context of shared-control powered wheelchairs in which adaptive assistance is provided to individuals who otherwise might not possess the fine motor control necessary to handle potentially challenging activities, such as doorway traversal. In this context it is important to go beyond the 2D laser scanner for open doorway detection, for both safety reasons as well as opportunities for improved shared-control behavior development. We evaluate the doorway detection by systematically testing the performance on several doors and door configurations from varied view points, using point clouds generated by a Microsoft Kinect.