Evaluating Time-to-Collision Models and Steering Feasibility for Autonomous Emergency Braking

An Autonomous Emergency Braking System (AEBS) relies on accurate Time-To-Collision (TTC) estimation for optimal performance. However, conventional AEB systems often suffer from delayed or missed detections (false negatives), where collisions are underestimated due to limited motion modelling while calculating TTC. The resulting impact is compounded due to slow response times from pneumatic systems in heavy commercial road vehicles, amplifying the need for a well-performing AEBS. Further, there exist situations where braking is initiated despite the existence of feasible avoidance paths that the driver intended to take (premature triggering/false positives). Motivated by the same, a comparative study of three TTC formulations is presented in this study. Further, to reduce undesired AEBS activations, a steering feasibility check is incorporated that evaluates lane-change with curvature and traffic constraints. The proposed framework is tested across Euro NCAP®inspired scenarios on a brake hardware-in-loop setup, and the constant acceleration-constant heading algorithm is seen to effectively capture both longitudinal and lateral dynamics while avoiding collisions in all test cases.