An Optimization Procedure to Maximize the Traction of a New Power Wheelchair

Alexandra N. Jefferds BS, Jonathan L. Pearlman PhD, and Rory A. Cooper PhD
Human Engineering Research Laboratories
Pittsburgh, PA 15206

ABSTRACT

We developed an optimization procedure to maximize the traction of the Hybrid Power Operated Vehicle (HyPoV) while enforcing stability and turning constraints according to Medicare’s Group 2 Power Mobility Device Codes. The HyPoV was well-received when tested in the US and India, and our goal was to optimize the frame geometry before building a third-generation prototype. Because the HyPoV’s casters are suspended on leaf springs for rough-terrain maneuverability, the GRF and tipping angle calculations require a model that is more complex than existing ones. We developed a parameterized mathematical model of the HyPoV and used two optimization procedures to calculate the HyPoV’s tipping angle, turning radius, drive wheel ground reaction force, throughout a relevant range of geometric changes. The first procedure was Matlab’s fmincon (a stepwise nonlinear programming approach) with GRF as the objective function, and the second involved the calculation of the GRF in a 100-point solution space.  We optimized the following design parameters (measured with respect to the front wheels): d (position of the drive wheel), s (end of swing arm), p (the linkage pivot), and c (casters). Ultimately, the latter two points were held constant during the optimization to speed up calculations. We had difficulty converging on a solution using the nonlinear programming method. However, the solution space method allowed us to identify an optimized geometry that predicted a 7% increase in ground reaction force (over the current prototype) while enforcing relevant turning and stability constraints. These results demonstrate that it was possible to learn about a mechanical system using our mathematical approach. Nonlinear programming could handle four dimensions more efficiently, and we plan to improve our implementation of this method in the near future.

KEYWORDS

Wheelchair design, wheelchair modeling, electric powered wheelchair, design optimization

ACKNOWLEDGEMENTS

This study was funded by the National Science Foundation grant # EEC 0552351. Additional thanks to Hongwu Wang.

Author Contact Information:

Alexandra Jefferds BS, University of Pittsburgh, Human Engineering Research Laboratories, Pittsburgh, PA, 15206, Office Phone (412) 365-4858, EMAIL: jefferdsa@herlpitt.org