Acquisition of Association Rules between a Situation and an Operation of Driving Behavior by using Time Series Clustering based on Coupled-GP-HSMM

With the progress of information and communication technology, it has become easier to observe and store a large amount of digital data. In the field of automobiles, various sensors such as Controller Area Network, GPS sensor, and in-vehicle camera are mounted in vehicles, and it is possible to collect location, operation, and surrounding environment information while driving. The collected driving behavior data can be used for accident analysis and driver’s behavior analysis. However, the driving behavior data is an extremely large amount of data, and it is not easy to manually extract the data of a specific driving scene which can analyze accident and driver’s behavior from a series of driving behavior data.In this paper, we propose a time series clustering method for driving behavior data. In this method, driving behavior data is divided into operation data and situation data. Here, the operation data expresses the accelerator amount, the brake amount, and the steering wheel angle, and the situation data expresses the vehicle behavior and surrounding environmental information. By applying Coupled-GP-HSMM, automatic segmentation and clustering of situation and operation data, and extraction of association rules between situation and operation are realized. In this paper, the effectiveness of the proposed method is evaluated by using the vehicle speed as situation data and the brake amount as operation data through the experiments.