A comparison between smartphone sensors and bespoke sensor devices for wheelchair accessibility studies

With the emergence of cheap sensors, there has been a proliferation of novel applications aimed at providing personalized services - from location and routing to daily energy consumption. Most existing applications are aimed at the able-bodied; however, Internet of Things technologies bring the potential to revolutionize the lives of the whole of our population. In this paper, our application area lies in supporting wheelchair users to navigate around the built environment in a way that best suits their disability. This is founded on the availability of good sensor data, and there are essentially two approaches to gathering this: (i) use bespoke sensors, with tight control over exactly what data is gathered and when; (ii) repurpose smartphone sensors to gather the necessary data. This represents a potential tradeoff between accuracy and easy availability. In this paper, we examine the differences in the data quality obtained from different kinds of devices and demonstrate that the smartphones can give a comparable data quality to bespoke devices. We also investigate how, and to what extent, the holding or mounting of the smartphone affects the measurements.