Towards a Safe, Low-Cost, Intelligent Wheelchair (2009)
Unlike most other robots, autonomous personal transports must be designed with a passenger user in mind. This paper examines the integration of three necessary technologies for a robotic transport—in particular, a robotic wheelchair. First, local motion to a nearby goal pose needs to be safe and comfortable for the human passenger. Second, 3D overhangs, drop-offs, steep inclines, and stairs (in addition to pedestrians and walls) need to be accurately modeled and avoided, while curb cuts, drivable ramps, and flat ground should be seen as traversable. Third, the spatial representation of the robot should facilitate infrequent requests for human directions and allow “natural” directional commands. Furthermore, the sensorimotor system that facilitates spatial reasoning, planning, and motion needs to be cost efficient. As a result, our goal is to create a system that ultimately uses inexpensive wheel encoders and off-the-shelf stereo cameras. In this paper, we overview the three technologies listed above. We then discuss the successes and the current failures of the integration task, both of which motivate future work.
In Workshop on Planning, Perception and Navigation for Intelligent Vehicles (PPNIV) 2009.

Patrick Beeson Postdoctoral Alumni pbeeson [at] traclabs com
Shilpa Gulati Ph.D. Alumni gulati [at] mail utexas edu
Benjamin Kuipers Formerly affiliated Faculty kuipers [at] cs utexas edu
Aniket Murarka Ph.D. Alumni aniket [at] cs utexas edu