We envision a future where partial-autonomy in powered wheelchairs will be the standard: that when a person is being fit for a wheelchair by a therapist, a variety of autonomy options will be available, just like today a variety of seating options are available.
Robotics autonomy can help with obstacle avoidance, navigation, route planning and spatially-constrained maneuvers. We are developing a shared-autonomy wheelchair that consists of modular software and hardware components, where the machine automation furthermore is customized to the physical needs and personal preferences of the user [6,11]. Our system also prioritizes simple integration with existing commercial chairs and control interfaces---to mitigate costs not covered by insurance and thus accelerate adoption by users.
For this platform, we have developed perception algorithms able to detect doorways , inclines and drop-offs  and docking locations . We currently are conducting a large-scale study with high-level spinal cord injured volunteers that evaluates multiple control-sharing paradigms and control interfaces.
A recent collaboration with Innovative Design Labs aims to transition the control-sharing components of our system in particular to a commercial product. Funding Source: National Institutes of Health (NIH/SBIR R43-HD085317).
Doorway Detection with Pose Estimation 
Docking Location Identification with Pose Information 
Autonomous Doorway Traversal with Obstacles
Autonomous Docking in Constrained Spaces
© argallab 2016