RERC: Design and Development D-6

D-6: Enhanced Controls for Powered Wheelchairs

Task Leader: Rory A. Cooper, PhD (rehabilitation engineer, consumer)

Co-Investigators: Cameron Riviere, PhD (engineer); Michael L. Boninger, MD (engineer, physician)

Duration/Staging of Task: This development task will be active in two phases; months 0-24, and months 37-60 of the 60-month RERC cycle, first phase commencing January 1, 1999.

Rationale/Task Summary
This research will make electric-powered wheelchairs safer and allow improved control. Most wheelchair manufacturers have not been able to adequately address the problems described in this task. The design and development strengths of the RERC will make it possible for manufacturers to integrate the findings of this study into future products. This will make manufactured products perform better to the benefit of consumers and clinicians. Clinicians will have better tools for tuning electric-powered wheelchairs, and consumers will have better mobility. Possibly, some consumers, who cannot independently operate an electric-powered wheelchair, will be able to do so with the technologies proposed.
Micro-controllers improve control not only by permitting repeatable tuning and control of certain system parameters, but also by allowing implementation of dynamic control algorithms (Cooper, 1995b). The most common control variable is speed. A typical speed control algorithm uses tracking, or contour, control to have the wheelchair automatically follow the desired speed profile set by the user, by varying the input device, regardless of terrain or slope. Hence the wheelchair/rider system will move at the same speed up an incline, as it will down an incline for the same user desired speed. The same is true when comparing a smooth surface to a rough surface. Speed feedback is done with an optical encoder in quadrature for direction, or by using the back emf of the motors. Heading is derived from using either tachometers or the back emf from both motors. The vast majority of EPW use the back emf, because of reduced cost and greater simplicity (Cooper, 1995b).

Design of development activities

Objectives

This development task will investigate the following assumptions:

1. Control of electric-powered wheelchairs can be improved by accounting for the uncertainty in the back electromotive force (emf) signal at low speeds.
2. The use of a jump-linear control system will provide better control while driving in reverse than current algorithms on commercial electric-powered wheelchairs.
3. Tuning the compliance of the joystick will provide improved control while driving over obstacles.
4. The addition of signal processing algorithms at the joystick can improve electric-powered wheelchair control.

Progress Report (12/31/99)

Surface mount components for the isometric/variable compliance joystick have been obtained. A four layer printed circuit board is under construction. IRB application for clinical trials is in preparation.

Methods and Technologies Proposed

This study will be conducted in to two phases: (1) the first phase will be to analyze the noise in the back emf signal and to develop modern control algorithms; (2) the second phase will be to develop a prototype adjustable compliance joystick and develop algorithms for this interface. Phase one will be used to test assumptions (1) and (2). Phase two will be used to test assumptions (3) and (4).

Publications/Reports

"Designing a Variable Compliance Joystick for Control Interface Research" Spaeth, DM. In: RESNA '99- Spotlight on Technology; 1999, Long Beach, CA. Washington DC: RESNA Press, 1999.

"Variable Compliance Joystick For Control Interface Research",Spaeth, DM; Cooper,RA. In: Proceedings of the 21st Annual International Conference of the IEEE Engineering in Medicine and Biology Society; 1999, Atlanta, GA.

Progress Report (12/31/00)

A working, commercial grade version of the force sensing joystick was completed during year 2000. Surface mount PC boards arrived in May and the Joystick became operational in August. Serial data ports (X and Y axis 10 bit, 100 hertz) were added during the fall. The Human Engineering Research Laboratories are currently completing ten additional joysticks from the initial run of PC boards. The ten new units will feature an improved enclosure design with waterproof and dust proof seals and space to accommodate a complete set of operator controls. Consumer testing is scheduled for Summer 2001.

photo of a joystick, parts, and tools

Progress Report (12/31/01)

At the close of year 2000, a single force-sensing joystick (FSJ) with redesigned printed circuit boards had been assembled and tested. A software program written in C++ was completed and allows a PC computer to record the force data being transmitted by the FSJ.
During the first two quarters of 2001, ten additional FSJ's were constructed using the new PC boards and surface mount electronic components. An all-metal enclosure was designed and machined with Computer Numerical Control (CNC) machine tools. The completed metal enclosure was the same size as those used with commercial joysticks. (Figure 1)

Figure 1. Comparison of one of the ten new HERL isometric joysticks to a standard position sensing joystick. Enclosure size and user controls are almost identical.

A full complement of user controls was provided including speed control, horn, power switch and battery status bar graph. Interface cables for chair control and data recording pass through a sealed flange at the rear of the new enclosure. A third PC board was created to support the user controls and provide space for future expansion. All ten units have been tested driving electric powered wheelchairs (EPW).
During the 3rd and 4th quarters of 2001, work began to interface the FSJ to personal computers for data collection during EPW driving studies and as a mouse cursor controller to support studies of virtual wheelchair driving and computer access (Figure 2).

Figure 2. Force Sensing Joystick interfaced to PC computer. Setup screen allows the threshold and response rate of the cursor to be adjusted

An evaluation station is under construction, which will allow subjects to use the FSJ while sitting stationary in their own wheelchair. This station will allow the FSJ to be accurately positioned at a location most conducive for operator control without mounting the FSJ directly to the subjects' wheelchair frame. The station has been designed for quick disassembly so that it may be transported to sites in the community for data collection. A C++ program is under development to translate the FSJ serial data into mouse cursor format. This will allow the FSJ to act both as a wheelchair control and as a mouse input device.
Two study protocols for the FSJ have been updated and reapproved by the Veterans Admin-istration Human Studies Research Board and the University of Pittsburgh IRB. The first study will compare the FSJ and conventional movement sensing joysticks for task of wheelchair driving using a modified Fitts' law paradigm. In this study subjects will drive an EPW ten to thirty feet and attempt to stop within the perimeter of a black target circle.

Figure 3. Optical sensor (one of four) under left foot rest; communicates with a laptop computer on rear of chair.

Four recently completed optical sensors mounted on the corners of the wheelchair (Figure 3 will detect whether the subject is able to stop within the target circle or rolled beyond the edge. In order to collect hand movement data from a conventional position-sensing joystick, a micro controller card in a small aluminum enclosure was designed to perform A/D conversion of the joystick signals and transmit the data in RS232 serial format. (Figure 4).

Figure 4. Modifications to position sensing joystick to support data collection.

Figure 5. A custom fitted aluminum tray allows a laptop computer to record data while the chair is in motion.

A specially constructed padded tray (Figure 5), fabricated of 1/4" thick aluminum plate mounts on the rear of the EPW and provides collision to the laptop computer while it records data during driving trials.
A full journal article comparing real vs. virtual EPW driving was completed by Cooper et al in the closing weeks of 2001 and is currently under review by Medical Engineering and Physics (See Publications section for Abstract)
A second upcoming study will further quantify the function of the FSJ as a cursor/virtual wheelchair control. In this study the subjects will operate the FSJ from the evaluation station described above and use it to direct a mouse cursor to onscreen targets. Different ratios of joystick compliance will be tested. Data collection for these studies is planned for the 1st and 2nd quarters of 2002.

Progress Report 12/31/02

The isometric joystick with two custom control algorithms developed for task D-6 was clinically tested during the second quarter of 2002. Thirteen subjects with significant disabilities and daily users of powered mobility were recruited and tested; performance data was collected. Inputs: construction of isometric joystick with custom control algorithms, C++ data collection software custom written for this project to collect data. Optical instrumentation was designed and fabricated to detect target acquisition. IRB approvals were obtained from the Department of Veterans Affairs and the University of Pittsburgh. Processes: Thirteen subjects were recruited and tested at the Human Engineering Research Laboratories, each subject drove an instrumented chair from a common starting position to nine circular targets (black vinyl disks 155 cm in diameter taped to floor). Each joystick algorithm, (two custom algorithms on an isometric joystick and a standard conventional joystick) was randomly presented three times for each of the nine targets, 81 trials total. Data was collected for reaction time, movement time and whether the subject successfully halted the chair within the circle. Data was also collected on the force generated by the subjects’ hand (12 bits resolution, 83 Hz) while the chair was in transit. Outputs: The data analysis revealed that the advanced algorithm employed on the isometric joystick provided wheelchair control performance statistically indistinguishable from a conventional movement sensing joystick (MSJ); despite the fact that all the subjects had years of experience with an MSJ and were using the isometric joystick for the first time during the study. A less sophisticated square template algorithm on the isometric joystick yielded significantly slower movement times than the MSJ.

Publications

Spaeth D.M.; Cooper,R.A.; Ammer BA; Guo SG; Wong C; Boninger ML, "The Benefits of Commercial Quality Prototyping - a Case Study of an Isometric Joystick", Conference Proceedings -Intellectual Property in the VA: Changes, Challenges & Collaborations, Office of Research and Development Technology Transfer Program and VA Rehabilitation Research & Development Service, April 2001

Spaeth DM; Cooper RA; Guo S; Wong C, "Using Quadrature Emulation to Connect Proportional Controls to Personal Computers Through a Standard Mouse", Conference Proceedings RESNA Annual Conference, Reno, Nevada June, 2001.

Cooper RA, Jones DK, Spaeth DM, Boninger ML, Fitzgerald SG, Guo S "Comparison of Virtual and Real Electric Powered Wheelchair Driving Using a Position Sensing Joystick and an Isometric Joystick, In Review, Medical Engineering and Physics, 2002.
Abstract - There are limited interface options for electric powered wheelchairs which results in the inability of some individuals to drive independently. In addition, the development of new interface technologies will necessitate the development of alternative training methods. This study compares a conventional position sensing joystick to a novel isometric joystick during driving task in a virtual environment and a real environment. The results revealed that their were few differences in task completion time and root-mean-square error (RMSE) between the two types of joysticks. There were significant correlations between the RMSE in the virtual environment and the real environment for both types of joysticks. The data indicate that performance in the virtual environment was representative of driving ability in the real environment, and the isometric joystick performed comparably to the position sensing joystick.