At a recent meeting, there was an interesting discussion on the topic of wheelchairs. My own brother uses a manual wheelchair, and my mother uses a motorized one, so this topic is of particular interest to me.
The current design of wheelchairs remains fairly unsophisticated. The motorized ones are basically a sort of low speed go-kart. The manual ones probably haven't changed much over the past century. Given the ramp-up in maker technology and inexpensive electronics and sensors, surely there are various ways to improve the experience of wheelchair users.
Let's consider some challenges a typical wheelchair user might experience:
The standard joystick-type controller seems intuitive for moving in different directions. However, those with hand strength issues or certain neurological conditions might have difficulty with fine course adjustments.
Add to this the challenge of navigating through narrow doorways or around obstacles. Especially with motorized chairs, it's just too easy to accidentally graze a door frame, or leave an occasional divot in the wall when backing or turning.
The cost of motorized wheelchairs can sometimes be comparable to that of an automobile. What if it were possible to retrofit existing manual wheelchairs with compact motors and batteries, for relatively short distances? Advances such as supercapacitors could make this a practical reality.
So, how to address these challenges? Here are a few ideas I've come up with so far:
The joystick-type directional controller is not remarkably different from any other kind of user interface, such as a mouse.
What about using a mouse touchpad as a wheelchair controller? Alternatively, there are low-cost magnetometers which can sense the angle of a nearby magnet, perhaps embedded in a ring, to provide directional input without the need for gripping force. The maneuvering sensitivity could be adjusted automatically based on motor speed - for example, it might be helpful to reduce sensitivity at higher speeds in order to prevent oversteering.
Most makers have seen the inexpensive ultrasonic sensors used for detecting obstacles. Recently I came across a wearable device which blind persons could use to detect obstacles in their path. These sensors could also be used as proximity sensors when backing up in a wheel chair, with perhaps one at each corner, angled outward, with one in the middle facing straight back. These sensors apparently only gave an angular detection range of about 15 degrees, so multiple sensors might be needed for practical use. Fortunately, these sensors only cost about two dollars apiece, so incorporating an array of these would not be too expensive. They could be inset into the frame of the chair, for example.
Alternatively, computer vision software such as the Open Source "OpenCV" can run on mobile Linux platforms such as the Raspberry Pi. This is much more programming than with sonar devices, but could potentially become much more flexible.
For example, a pair of cameras could provide distance cues based on stereoscopic scene analysis over a wide angle.
The Xbox Kinect uses a pattern of projected dots to calculate distances and movement. There is a developer kit to use a Kinect with a PC. This method could make use of a Windows-based tablet computer to assist in sensing hidden obstacles, or even for vision impaired persons.
The Leap Motion controller is a novel device that can detect fine hand movements in real time, replacing the mouse in specific situations such as gripping a virtual object in Augmented Reality. This technology might be adaptable to provide detailed proximity sensing as well, via the Leap Motion API.
Approaching narrow door frames or other constrained areas can require fine hand movements, which can be challenging for some wheelchair users. Perhaps we can borrow a few ideas from aircraft carrier landing systems!
As I recall from reading somewhere, aircraft carrier pilots use radar or similar technology to guide the aircraft to a safe landing under various conditions. We probably don't require that level of complexity to navigate at low speeds in close quarters. For example, a pair of ultrasonic sensors (as described above) could be used to monitor clearances ahead and slightly to each side when approaching a doorway.
If the distances remain close to equal during the approach, the the chair is centered between the door posts. If a significant difference is detected, then the user could receive some form of feedback to make the necessary steering correction. It may even be possible to kick in automatic feedback to adjust motor speeds as needed to remain centered.
On the subject of instant user feedback, existing visual or auditory signals have their drawbacks. Users should not have to keep looking at a screen while trying to maneuver. Audio feedback might be a better solution, although speakers or earphones can become annoying or interfere with normal spoken communication.
Since the user is already seated, an array of cellphone vibration buzzers may be incorporated into the backrest. The position of these devices can provide discreet situational feedback based on buzz pulse frequency and/or intensity, without interfering with vision or hearing. Even Helen Keller could have used that!