J R Mackenzie

A robust estimation of the effects of motorcycle autonomous emergency braking (MAEB) based on in-depth crashes in Australia

ABSTRACT Objective: Autonomous emergency braking (AEB) is a safety system that detects imminent forward collisions and reacts by slowing down the host vehicle without any action from the driver. AEB effectiveness in avoiding and mitigating real-world crashes has recently been demonstrated. Research suggests that a translation of AEB to powered 2-wheelers could also be beneficial. Previous studies have estimated the effects of a motorcycle AEB system (MAEB) via computer simulations. Though effects of MAEB were computed for motorcycle crashes derived from in-depth crash investigation, there may be some inaccuracies due to limitations of postcrash investigation (e.g., inaccuracies in preimpact velocity of the motorcycle). Furthermore, ideal MAEB technology was assumed, which may lead to overestimation of the benefits. This study sought to evaluate the sensitivity of the simulations to variations in reconstructed crash cases and the capacity of the MAEB system in order to provide a more robust estimation of MAEB effects. Methods: First, a comprehensive classification of accidents was used to identify scenarios in which MAEB was likely to apply, and representative crash cases from those available for this study were populated for each crash scenario. Second, 100 variant cases were generated by randomly varying a set of simulation parameters with given normal distributions around the baseline values. Variants reflected uncertainties in the original data. Third, the effects of MAEB were estimated in terms of the difference in the impact speed of the host motorcycle with and without the system via computer simulations of each variant case. Simulations were repeated assuming both an idealized and a realistic MAEB system. For each crash case, the results in the baseline case and in the variants were compared. A total of 36 crash cases representing 11 common crash scenarios were selected from 3 Australian in-depth data sets: 12 cases from New South Wales, 13 cases from Victoria, and 11 cases from South Australia. Results: The reduction in impact speed elicited by MAEB in the baseline cases ranged from 2.8 to 10.0 km/h. The baseline cases over- or underestimated the mean impact speed reduction of the variant cases by up to 20%. Constraints imposed by simulating more realistic capabilities for an MAEB system produced a decrease in the estimated impact speed reduction of up to 14% (mean 5%) compared to an idealized system. Conclusions: The small difference between the baseline and variant case results demonstrates that the potential effects of MAEB computed from the cases described in in-depth crash reports are typically a good approximation, despite limitations of postcrash investigation. Furthermore, given that MAEB intervenes very close to the point of impact, limitations of the currently available technologies were not found to have a dramatic influence on the effects of the system.

A trial of retrofitted advisory collision avoidance technology in government fleet vehicles

In-vehicle collision avoidance technology (CAT) has the potential to prevent crash involvement. In 2015, Transport for New South Wales undertook a trial of a Mobileye 560 CAT system that was installed in 34 government fleet vehicles for a period of seven months. The system provided headway monitoring, lane departure, forward collision and pedestrian collision warnings, using audio and visual alerts. The purpose of the trial was to determine whether the technology could change the driving behaviour of fleet vehicle drivers and improve their safety. The evaluation consisted of three components: (1) analysis of objective data to examine effects of the technology on driving behaviour, (2) analysis of video footage taken from a sample of the vehicles to examine driving circumstances that trigger headway monitoring and forward collision warnings, and (3) a survey completed by 122 of the 199 individuals who drove the trial vehicles to examine experiences with, and attitudes to, the technology. Analysis of the objective data found that the system resulted in changes in behaviour with increased headway and improved lane keeping, but that these improvements dissipated once the warning alerts were switched off. Therefore, the system is capable of altering behaviour but only when it is actively providing alerts. In-vehicle video footage revealed that over a quarter of forward collision warnings were false alarms, in which a warning event was triggered despite there being no vehicle travelling ahead. The surveyed drivers recognised that the system could improve safety but most did not wish to use it themselves as they found it to be distracting and felt that it would not prevent them from having a crash. The results demonstrate that collision avoidance technology can improve driving behaviour but drivers may need to be educated about the potential benefits for their driving in order to accept the technology.