Estimation of wheelchair states during movement using WELL-SphERE for evaluation of power wheelchair safety

To comprehensively evaluate the usability and safety of a power wheelchair (PWC), monitoring multimodal data related to the PWC in a real environment is crucial. In most studies exploring actual wheelchair conditions, modification of PWCs has been required. Especially modification of controlling circuits aiming for measurement of joystick operation may lead to controller malfunction and thus increase safety risk. It is essential, therefore, to ensure the safety of PWC users during experiments so that they can measure PWC-related data with their own wheelchairs. To achieve this aim, we developed a recording device that is easily installed on PWCs without any electronic modifications. The device, called a “WELL-SphERE,” has sensing units that can be attached to PWCs a data management unit that can store and transfer measurement data. Here, we focused on joystick operation logged by the system. Seven participants were pre-tested to examine the characteristics of logged operations during runs through four test courses. Subsequently, all participants completed a questionnaire regarding the difficulty of the test courses. From these results, we classified the logged operations into four categories of “wheelchair states.” Two participants-a novice driver and a mature driver-were also evaluated to verify the accuracy of the estimated wheelchair states. The accuracies of the estimates by the mature and novice driver were 98.8% and 89.0%, respectively. The wheelchair states for both participants showed characteristic patterns. Therefore, the wheelchair states estimated with the data logged using WELL-SphERE are valid indicators of the wheelchair conditions during movement.