Citation

Robinson MD, Beal CE, Brennan SN. Accid. Anal. Prev. 2020; 141: e105492.

Affiliation

The Pennsylvania State University, 227 Reber Building, University Park, PA 16802, United States.

Copyright

(Copyright © 2020, Elsevier Publishing)

DOI

10.1016/j.aap.2020.105492

PMID

32305618

Abstract

Pedestrian avoidance algorithms often tacitly assume that the maneuver which minimizes collisions will also be the safest maneuver. This work shows that this is not always the case when considering pedestrian fatalities. Given the unavoidable uncertainty in vehicle motion, environmental parameters, and pedestrian behavior, emergency avoidance maneuvers often involve some chance of a collision. Maneuvers that aim to keep the vehicle as far away from the pedestrian as possible will theoretically minimize collisions; but if this strategy is followed and a collision occurs nonetheless, it will often be at a higher speed than would occur with alternative strategies. This is a result of the tires' friction ellipse which enforces a constraint between steering versus braking; for collision avoidance, braking must be reduced if pedestrian clearance is to be maximized. This work shows that in some common pedestrian collision situations, the net effect of this increase in vehicle speed for pure avoidance offsets the benefits of reducing collisions. Pedestrians, if hit, would be hit at higher speeds leading to a net reduction in pedestrian survivability for collision-minimizing maneuvers. First, this trend is demonstrated and explained using a simplified point-mass model of a vehicle, which is then verified with a higher-fidelity vehicle model as well as experimental maneuvers with an instrumented vehicle. While real accidents involve dozens of important parameters, this research provides a general framework for an under-recognized effect under certain common conditions. The implication of this finding suggests that future research in pedestrian avoidance should consider fatality minimization as an alternative objective to collision minimization.

Copyright © 2020. Published by Elsevier Ltd.

Language: en