Radar-based target identification and tracking on a curved road

Existing target determination methods for radar only work properly on straight roads, so the application of radar in driver assistance systems, such as adaptive cruise control (ACC), is limited to detecting target vehicles directly in front of the host vehicle. Preceding vehicles around a curve cannot be detected by current radar systems. This paper describes a new algorithm for radar-based target vehicle identification and tracking on curved roads for ACC applications. The key problems are the distinction between target curve entry/exit and lane change and the identification of the host-lane preceding vehicle from the adjacent-lane vehicle while driving through a curve. In order to make a more accurate distinction between curve entry/exit and lane change of the target vehicle, the goodness of fit (GOF) corresponding to these two scenarios is found from the correlation coefficient and by introducing additional distinct criteria that have clear physical meanings. Furthermore, the deviation of the measured lateral distance from a theoretical value is used to estimate whether the preceding vehicle is in the host or adjacent lane. Tests show that this method can identify the position of the target vehicle on a curved road with high reliability.