Risk driving behaviors detection using pressure cushion

With the increasing frequency of traffic accidents, traffic safety has attracted attention of the researchers. Most of the traffic accidents are related to the driver’s risky behavior or some improper driving habits, such as leaning against the window/door, picking up things, or looking backwards when driving at high speed. In this paper, to detect such risky behaviors, we propose a decision tree for classification that recognizes four kinds of driving behaviors: normal driving, looking backwards, leaning against the window and picking up things. A time series of pressure data were measured from a mat with 2 × 2 pressure sensors which are distributed on the driver seat. Regarding the preprocessing phase, a digital filter is used for noise reduction. Results show that our method can achieve an average recognition rate of 88.25%.