Run-Off-Road Training Program, a Reflection

My Experience

I have experience driving several different make and model vehicles in normal and performance driving environments. I grew up working in an automotive performance shop, where we would modify customers vehicles, then test and tune the changes made to get the best performance out of the car. This experience is relevant to the publication being reviewed in this report because I understand what it is like to lose traction and control of various size, weight, and drive types of vehicles. I drive a Subaru Forester, which is all wheel drive and can perform well in a run-off-road event. Front wheel drive and rear wheel drive vehicles I have driven in the past are not as adept at regaining traction and control after a run-off-road event. I have experienced the consequences of a run-off-road event, spinning out and nearly being involved in a serious accident. I understand the importance of this problem, and think it is important to not only prevent these events from occurring in the first place, but also to explore safety systems or training solutions that make drivers better prepared to maneuver out of a run-off-road event.

Purpose and Expectations

Before reading the publication [1], I want to share what I think the purpose of the paper is and what my expectations are of the report. The title of the paper “A video based run-off-road training program with practice and evaluation in a simulator” indicates that the purpose is to discuss the effectiveness of training and practice on individuals ability to recover from a run-off-road event. I expect that there will be information shared about the statistical importance of decreasing fatalities and injuries in run-off-road events that supports the need for this kind of research. I expect that the paper will present the methodology behind the study to prove the efficacy of the training and practice programs on driver’s performance. There should be a discussion on the simulation setup and how closely it relates to real world scenarios. The paper should provide sufficient statistical evidence based on results from the simulation before and after the training, as well as qualitative data from the drivers in the form of a survey on how well they think they improved as a result of the program. I would like to finish the paper with a clear image of how this program can be expanded and shared with other drivers to improve the overall safety of run-off-road events.

Reflections

Immediately I was given the statistical significance as to why run-off-road events are a significant cause of crashes in the U.S. where in 2011 60% of all single-vehicle crashes were run-off-road related events. A digression, I wonder what impact autonomous vehicles would have on reducing run-off-road events, and if we can learn from our own mistakes to engineer a more robust path planning algorithm for future autonomous vehicles. Specific factors attributing to run-off-road events mentioned in the paper that are still cause for concern in autonomous vehicles are sudden obstacle avoidance and the reaction immediately following roadway departure. Countermeasures that currently are in place to reduce the severity of run-off-road events are not able to be implemented on all roads, but when the time comes to replace sections of roadway it should be a requirement that at least some of the most effective solutions be put in place. I thought it was interesting that the paper mentions electronic stability control (ESC) because in my experience it has only ever caused me more problems as an experienced driver than it solved. I was making an exit on the highway after coming out of a left-hand bend, and at the speed I was traveling, my ESC became active as I tried to make the right-hand exit. As a result, my steering wheel tried to over correct to push my back to the left. I nearly was steered into the impact attenuator; as a result, I modified my vehicles electronics system to completely disable this system as it was doing more harm than good. I believe there should be more training and practice invested in drivers to not become reliant on vehicle technologies that attempt to make decisions about the steering direction without more information about the environment and other factors. The methods section begins by providing an excellent overview of the participants, putting the study into perspective. The training video was created by using material from industry leaders in automotive safety, this was very well thought out and researched, however, maybe better information existed elsewhere that could have also been included to increase the efficacy of the video for training. The simulator description does meet my expectations, it provides a clear overview of the system but does not comment on the differences that the simulator has with respect to a real vehicle. Would a more realistic simulator improve drivers’ performance in recovering from run-off-road events? I felt that the simulated environments provided a concise set of challenging run-off-road scenarios where speed, lane width, and regions were considered appropriately. The test procedure includes details surrounding how they evaluated the qualitative difficulty of the recovery using the Likert scale. I wondered how effective this scale was at capturing the true opinions of the participants, however it is time efficient and is proven to work over other alternatives. The numerical evaluation considers three main metrics for analysis: maximum lateral error, the amount of steering angle change in a 4 second window, and the yaw rate and sideslip during the recovery process. I thought that using the ratio between the maximum measured sideslip and yaw rates for determining stability was a good indication of the driver’s performance during recovery. It does not entirely take into account the vehicle’s dynamics in terms of roll or pitch which, depending on the size or weight of the vehicle, could lead to very different definitions of what it would mean to have a stable recovery. The mental demand metric does not show much value since it is derived from a subjective survey and an average of responses.

Figure 1. Two vehicles with the same lateral offset but different yaw angles.

The results shown in figure 3 show that the treatment group improved over the control group in every metric except lateral error for all roadways [1]. This makes sense because it is possible that a small lateral offset can be achieved with stable and unstable yaw rates however the likely hood of a safe recovery is greater in the event that both a small lateral offset and yaw rate were observed. This scenario is depicted above in figure 1, where both cars have the same lateral offset but different yaw angles.

Next Steps Moving Forward

This publication is effective in providing clear evidence for the efficacy of driver training on the increased performance of drivers recovering from run-off-road events. It does a good job in deriving objective quantitative metrics based on the vehicles dynamics in evaluating the drivers pre and post training performance. The survey results did not provide much insight into the drivers’ cognitive demand and this aspect of the study seems to fall short. Further research into the drivers psychological and cognitive abilities as a result of training would be interesting on a neurological basis. Altogether I was satisfied with the results and found the paper did well in addressing the problem and providing evidence for an effective solution.

Conclusion

This paper did well to clearly introduce the importance of why the reader should care about solutions to reducing injuries and fatalities due to run-off-road events. The experimental setup was discussed in a concise manner and provided transparent information on what was done to simulate a real run-off-road scenario. The quantitative metrics used to determine the stability of the vehicle combined with the statistical analysis showed a clear improvement between the treatment and control groups before and after training. Altogether the paper could have done more to address the qualitative questions posed to the participants in the questionnaire, but this lack of evidence did not take away from the conclusion that training for run-off-road recovery greatly improves drivers ability to stabilize the vehicle. That conclusion on its own is enough to persuade readers that more should be done to train drivers and reduce the number of crashes caused by run-off-road events.

References

[1] Freeman, Paul, et al. “A Video Based Run-off-Road Training Program with Practice and Evaluation in a Simulator.” Accident Analysis & Prevention, vol. 82, 2015, pp. 6–6., doi:10.1016/j.aap.2015.05.008.