Introduction A baby learning to walk takes her first steps toward the waiting arms of her parents, who are watching carefully to make sure she doesn't fall. Children learning to ride bicycles start out with training wheels. Student drivers practice in dual-control cars, or on driving simulators. People learning how to pilot motorized wheelchairs, however, do not have a safe and stress-free environment to learn in. If you ask physical rehabilitation therapists how they teach patients to use wheelchair controls, you will be told that they put the patient into a chair, turn it on, and let the pilot "crash into walls until they get the hang of it". That is what I was told, by a physical therapist I was interviewing. She then asked if I could come up with a better solution -- one that would be less frustrating for a patient already undergoing a very frustrating rehabilitation process (motorized wheelchairs are usually used by stroke victims, or accident victims with spinal cord injuries). This is what inspired me to propose this project: an inexpensive, PC-based, "flight-simulator" for motorized wheelchairs. Background For the past year, I have been working in the Newman Lab as part of the Manus project, an ongoing project to develop a robotic system for physical therapy. The Manus device is a computer-driven robot "arm" that emulates the movements a physical therapist might make while working with a patient to re-develop arm and hand coordination. The robot arm holds the patient's own hand and arm, and either manipulates them according to a pre-set program, or records the patient's own movements for later analysis. There are several potential advantages in using this device. One advantage is the ability to use the patient's movements to drive a graphical goal and feedback system on a computer screen (In laymen's terms, a "video game"). As a student with background in both Cognitive Science and Graphic Design, it has been my job on the project to design and implement these games. The response to them from physical therapists has been extraordinarily enthusiastic. In the course of the Manus project, I have had the opportunity to interview several physical therapists, and to become somewhat familiar with the environment they work in. As I said in my introduction, the idea for my project came from one of these therapists. The Flight Simulator could therefore be considered an offshoot of the Manus project. The focus of that project, the Manus robot arm, will not be necessary for the "flight simulator", however. The flight simulator will consist of a translator box (described below) and another set of specifically designed video games. Industry Support, Existing Products In the course of researching the feasibility of this project, I have spoken to representatives of the country's two major wheelchair manufacturers. My first industry contact was with Everest & Jennings, the wheelchair company that supplies most of Massachusetts' rehabilitation facilities. My contacts at Everest & Jennings were very enthusiastic about the proposed flight simulator. A technical support engineer at that company has already provided me with technical specifications for their wheelchair controllers and control input devices, and has offered to give me used equipment with which to build a prototype (a typical wheel chair controller - the "brains" of the chair - costs about $2,500 new. A control input device, such as a joystick, costs about $250). He has also offered to answer technical questions as they arise. My second industry contact was with Invacare. This company is now marketing a computerized training system for wheelchairs on a limited basis. Invacare's system consists of a stationary dedicated wheelchair connected to a dedicated PC. The system lets the patient use the chair control input device to "move" through a driver's eye view of "real-life" situations. According to my contact at that company, this product, which is considered extremely expensive, is "not a huge seller." It is of interest, he said, only to high-level quadriplegics with substantial health insurance. (high-level quadriplegics do not have the use of their hands, and typically control wheelchairs using some form of mouth control). For most patients, he said, the therapists prefer to teach wheelchair control by trial and error (i.e., letting them crash into walls). Through my interviews with physical therapists, I have come to believe that the failure of Invacare's system is not due to dis-interest on the part of physical therapists, but rather to the high cost and limited appeal of their product. I also believe that providing a driver's eye view of the scene to be navigated is inappropriate for this application. Unless the system can also provide the appropriate vestibular and physical/inertial stimulation to the driver, the driver's eye view feature will ultimately cause more transition problems than it solves. I prefer to use my video games to teach basic control device operation (a substantial issue for sip-stick or headband users), and to let users become familiar with wheelchair driving issues such as "wall jamming" (it's impossible to turn a wheelchair that has run forward into a wall. The user must first back it up a couple of feet). From the technical discussions I have already had with the engineer at Everest & Jennings, I believe that it will be possible to create a prototype of the simulator within three months. Description (and Technology) The simulator system I envision will consist of the following: Hardware: 1 -- An inexpensive translator box that will link the output from the patient's own wheelchair control device to a joystick port on a PC. According to the technical support engineer at Everest & Jennings with whom I have discussed the project, the I/O for a standard motorized wheelchair joystick is similar to the I/O for a computer joystick (DC, low voltage). He agrees that the slight voltage change necessary could be performed "on the fly" using appropriate hardware. My goal for the prototype will be to create a simple piece of hardware that can perform the translation. This will be the only specialized hardware necessary to run the system. 2 -- An ordinary PC. Many physical therapists already use simple PC games for rehabilitation exercises. The games already available are typically mouse-driven "connect the dots" type games that help develop eye-hand coordination. It would not be necessary for a rehabilitation facility to purchase a dedicated PC in order to use my system. 3 -- A joystick port card. These can be added to any IBM-compatible PC for less than $50. 4 -- The patient's own wheelchair control device. According to my industry contacts, the control device is a modular unit that can be detached from the wheelchair controller. There are several kinds of control devices: "sip-stick" controls for high level quadriplegics, controls based on head bands for people who can move their necks, joysticks, etc. For the prototype, the wheelchair control device will be of the joystick type. It is my hope, however, that the system will be flexible enough to eventually be used with any kind of control device. Software: One module of the software will be a controller emulator, that will allow the games to be used at different levels of joystick sensitivity. Chair controllers are programmed to respond differently for different users. This module will emulate that programming, and may be of use to physical therapists in deciding how to program chairs for individual patients. The main module of the software will be a suite of video games. This project has a lot of potential for creating different kinds of games, of varying difficulty levels. Because many of the users of this system will be mentally intact (if physically challenged), at least some of the games should be intellectually interesting. In addition, the games should incorporate varying levels of difficulty, so that even the weakest users can feel they are "succeeding" (difficulty can be increased as mastery increases). I will have to do some more interviewing with physical therapists before I finalize my software designs, but some of the basic games I envision are as follows: 1 - A "doodling" game. This game will draw a cursor on an otherwise blank screen. Input from the control device will move the cursor. As a long time computer user, I know that being able to "just play" with new equipment is a very effective way to get a feel for its operation. This game probably wont' be used for very long by any one patient, but I think using it will increase the psychological comfort level of most people in the process of learning to use a motorized chair. I also think it may be a useful tool for physical therapists in determining the readiness of patients to begin wheelchair training. 2 - A maze game. This game will present a variety of mazes on the screen, which the patient will have to navigate through. The cursor in this game should exhibit standard wheelchair "quirks" such as wall jamming, wide turning radius and wide undercarriage - some of the paths in the maze that appear clear will actually be inaccessible because they are too narrow or require too sharp a turn. 3 - A "Donkey Kong" type game. This game will be similar to the maze game, but will show the cursor and obstacles from an edge-on perspective. Patients will learn to avoid ramps that are too steep, and will be penalized for going off curbs that do not have curb cuts. (If all this sounds frustrating, consider trying to do it the for the first time in a real-life situation, with no reset button.). Budget Below is a rough estimate of my expenses for the project: Living expenses $2,500 Hardware for translator box 500 C++ developers kit 500 Long distance phone calls and fax/fedex 200 Rental of a motorized wheel chair (I plan to drive one around for a week, so I can discover more driving issues) 500 Computer upgrade (bigger hard disk, math co-processor, better screen) or partial purchase price of new computer 800 The Wheelchair Flight Simulator Proposed by Amy Gorin P.O. Box 11, Somerville MA, 02144 Course 2A, Class of 1994