Driver Steering Override for Lane-Keeping Aid Using Computer-Aided Engineering

Abstract

The decreasing cost of radar and camera sensory technology has established advanced driver assistance systems (ADAS) to the standard equipment of modern vehicles. ADAS employ semi-autonomous interventions in longitudinal (e.g., adaptive cruise control, ACC) and lateral (e.g., lane-keeping aid, LKA) direction. ACC and LKA constitute the milestones of ADAS technology and have shown to correspondingly reduce driving effort and unintended lane-drifts. ADAS and driver share the vehicle control; this in-turn creates a nonfinite number of test cases that have to be tested-verified so-as to enable safe software (SW) release. Adjacent to ADAS, vehicle dynamics-related systems, such as the antilock braking system and the dynamic-stability and traction control, have also to undergo through an exhaustive SW verification. Distilling the previous statements, this paper cultivates the concept of virtual SW verification by evaluating Volvo's production and a concept driver steering override (DSO) for LKA systems. The DSO concept addresses the driver's interaction with the LKA system and adapts the intervention level. The driver's activation is quantified though the steering torque and road/vehicle information. Two different test-runs per DSO strategy are presented and the real-car (physical) testing results are reproduced in simulation. Both the real and postprocessed simulation results show: 1) the importance of the override strategy on the LKA benefits; and 2) the importance of computer-aided engineering for studying/designing ADAS.