2017 Volume 137 Issue 4 Pages 658-666
We propose a novel road monitoring system named YKOB (Your Kinetic Observation Bike) based on participatory sensing. YKOB collects acceleration signals using smartphones worn by cyclists, and analyzes the collected signals to investigate road surface condition. When a bicycle passes on a bump or a dimple, its wheels vibrate. The vibrations are transmitted to the smartphone via the bicycle frame or the cyclist body, and registered as acceleration signals. Conversely, by analyzing the acceleration signals we can estimate the road surface condition. There are mainly two research issues in this system. The first issue is that the acceleration registered at the smartphone includes cyclist motion signal as well as road surface signal. The second issue is that it is necessary to distinguish abnormality of road surface from artificial differences in level, such as a difference between streets and sidewalks. We developed a signal separation algorithm based on Independent Component Analysis (ICA) to solve the first issue. We also developed a bump classification algorithm using Real Mother Wavelet (RMW). These two proposed algorithms were evaluated with 640 trials in total of experimental data conducted by eight cyclists. The classification accuracy of 0.68 validates the simultaneous utilization of our proposed algorithms.