Consistent Threat Assessment in Rear-End Near-Crashes Using BTN and TTB Metrics, Road Information and Naturalistic Traffic Data

Artikel i vetenskaplig tidskrift, 2017

Rear-end crashes are one of the most frequent types of traffic accidents. As a response, in order to assist the drivers, Advanced Driver Assistance Systems (ADAS) are being developed; particularly, Collision Avoidance Systems (CAS). Current CAS algorithms are commonly based on very simplified models of the preceding vehicle motion, under the assumptions of constant velocity or con-stant acceleration. This leads to several issues in the pre-diction of the lead vehicle behavior, and at the end, in the evaluation of the potential collision hazard. This paper identifies and addresses a hazard overestimation issue occurring in events related to rear-end (near) crashes that take place at traffic junctions. The issue was identified by analyzing 121 events found within data of 14640 driving hours collected in the framework of the Eu-ropean Field Operational Test on Active Safety Systems (euroFOT) project. Aiming to solve this issue, in this work it is proposed an enhanced model including road information to predict the lead vehicle braking behavior (deceleration pattern) of the CAS algorithms. Then, the hazard estimation is han-dled by a consistent event detection strategy, which uses simultaneously two metrics: the Time-to-Brake (TTB), and the Brake-Threat-Number (BTN). The result is the Consistent Threat Assessment for Longitudinal Motion Al-gorithm (CTALMA). The applicability of this algorithm is illustrated by an in-depth analysis of three test examples. Obtained results show a better assessment of the collision potential hazard and consistent event detection avoiding false alarms in all the cases.