This content is not included in your SAE MOBILUS subscription, or you are not logged in.
Empirical Study of the Braking Performance of Pedestrian Autonomous Emergency Braking (P-AEB)
ISSN: 0148-7191, e-ISSN: 2688-3627
To be published on April 14, 2020 by SAE International in United States
Vehicle manufacturers are beginning to improve existing autonomous emergency braking (AEB) algorithms by pedestrian identification and avoidance capability. The Insurance Institute for Highway Safety (IIHS) has performed tests on eleven such vehicles; data is publicly available and was analyzed for this study. The IIHS tests were divided into three scenarios- 1) An adult pedestrian crossing a street on a path perpendicular to the travel line of vehicle, with a vehicle approach speed of 20 or 40 km/h, 2) a child pedestrian crossing a street from behind an obstruction on a path perpendicular to the travel line of a vehicle (approach speeds 20, 40 km/h), and 3) an adult pedestrian near the edge of a road in a path parallel to the travel path of a vehicle (approach speeds 40, 60 kph). An analysis was performed to compare Forward Collision Warning (FCW) engagement time, brake application time, and probability of impact across different manufacturers. It was observed that FCW on time for the 2019 Volvo XC40 lasted from 0.95 sec. - 2.36 sec., whereas for the 2018 BMW X1 the range was 0.08 sec. – 1.3 sec. The Volvo impacted the pedestrian dummy in only 2 out of 32 tests whereas the BMW impacted the dummy in all valid tests. The second objective of this study was to analyze the brake application characteristics of pedestrian AEB, and how they vary across various manufacturers. It was observed that most vehicles equipped with pedestrian AEB applied brakes using a two-phase strategy, consistent with existing studies of autonomous emergency braking. Vehicles exhibited an initial phase of gradual application (Phase 1), followed by a steady-state phase (Phase 2). The data was filtered and studied to understand the duration and magnitudes of Phases 1 and 2. Volvo test data showed that the maximum Phase 1 braking averaged 1g with a duration of about 0.33 sec. The BMW test data for Phase 1 braking showed an average maximum of 0.3-0.4g while the duration ranged from 0.3 sec – 0.5 sec. The initial ramp up in Phase 1 braking in the Volvos was observed to be much steeper compared to the BMW. The Volvo appears to be adjusting the magnitude of its Phase 2 steady state braking depending on the current velocity and distance from the pedestrian. In the lower velocity ranges, the magnitude of the steady state region averages 0.4g. In the higher velocity ranges, the vehicle tries to maintain a maximum braking effort of 1g.