Four-way joystick systems for steering, braking, and accelerating are sometimes used to adapt cars to drivers with severe physical disabilities. Previous research has revealed problems with such systems, such as interference between lateral and longitudinal control, lack of feedback and time lags, which make these systems difficult to use and hard to learn. The behaviour of the car is difficult to predict for the driver while driving such systems. A driving simulator experiment was conducted with the aim to investigate alternative joystick designs. Yet another objective was to contribute to the development of a method that could be used to evaluate vehicle adaptations in terms of how well the adaptation compensates for the driver’s disabilities. Thus, four different joystick designs were developed and tested by varying two factors: the degree of interference between lateral (steering) and longitudinal (speed) control (coupled/decoupled) and force feedback to the driver (passive/active). Time lags were made similar to what is found in conventional controls (steering wheel and pedals) in standard cars. It was expected that decoupling and active feedback would provide better control, less workload and make it easier to learn to drive with a joystick. Sixteen experienced drivers with a spinal cord injury at cervical level (C4–C7) participated in the experiment. However, they were all inexperienced with joystick driving. The drivers were all paralysed in their legs and the majority had a degraded function in their arms and hands. The driving task consisted of both driving on a rural road and a double lane change manoeuvre. All subjects drove with all four joystick systems. A cognitive systems engineering approach was used to argue for time based safety margin measures as a tool to determine which design was superior. However, there were rather few significant differences revealed between the four joystick designs. Partly this could have been a result from the composition of the driver group. Even if the participants were diagnostically homogenous they were functionally diverse which contributed to a large variation in data. Thus, the analysis was done both for the total group and for two separate groups of equal size depending on their arm and hand function. It was found that decoupling of lateral and longitudinal control at least partly provided better control and less workload. This was specifically true for with drivers with better hand and arm function. Active feedback was also experienced as positive together with the decoupled control by the same group of drivers and provided them with better control during the lane change manoeuvre. However, the drivers with less arm and hand function were in more favour of passive feedback and it seemed like the active feedback forces were even disturbing to them. It was concluded that the feedback forces should be individually adjusted. The reduction of time lags (compared to systems available on the market) contributed to make it very easy to learn to drive with the tested joysticks but active feedback and decoupling did not facilitate learning. Finally, it was concluded that a manoeuvre test should be included in an adaptation evaluation as it was considered to have a good potential to disclose insufficient adaption. (Author/publisher) This document is also available at http://www.vti.se/EPiBrowser/Publikationer%20-%20English/R506A.pdf
Samenvatting