REXASI-PRO tests smart wheelchair autonomy

This is the right kind of ambitious assistive-tech research: not “replace the user,” but make mobility safer when the environment gets messy. The hard part is less the AI demo and more making the system reliable, explainable, and affordable enough to survive daily life.

• –Shared-control navigation feels like the near-term winner; full autonomy is exciting, but user override and trust matter more in a wheelchair than in a lab robot
• –Adding room sensors and drone-based cameras improves awareness, but it also increases calibration burden, cost, and failure modes
• –Natural-language driving is a strong accessibility story, especially for users who can’t rely on fine joystick control, but it needs rock-solid confirmation and emergency-stop behavior
• –The biggest question is deployment, not capability: can this outperform a skilled wheelchair user in the real world without making the experience more fragile?
• –If this class of systems matures, the likely product shape is “intelligent assistance” first, “fully autonomous transit” much later