Abstract
In this paper, we propose novel approach to estimate traffic interactions with surrounding traffic participants. Recently, various advanced driver-assistance systems (ADAS) have been proposed. In these ADAS, however, it is not sufficiently considering the influence of own vehicle behavior. With a novel driver assistance system based on the traffic interactions, each vehicle keeps not only own vehicle but also surrounding space in safety and comfortable, such as, following-distance control system for reducing traffic jam. We estimate the interactions from the behavior data of the traffic participants using Bayesian filtering techniques. Efficiency of the interaction estimation is evaluated in simple traffic simulations, i.e., Cellular Automaton Model. In the simulated experiments, following-distance control system based on interaction estimation improves traffic flow 140% smoother than without the driving support. Development of the specific ADAS application based on traffic interaction and demonstrations of effectiveness using real-vehicles are important feature works.
Keywords: Interaction Estimation, ADAS, Driving Context