ExxonMobil makes significant natural gas discovery offshore Cyprus
NGO testifies biomethane should be part of NY state climate legislation

Rice, Texas Tech study finds automated cars and their operators fail to detect dangers

New research from Rice University and Texas Tech University has found that drivers often fail to spot hazards missed by automated vehicles, and it only gets worse the longer drivers ride in them. The study is published in the journal Human Factors.

The researchers examined the behavior of 60 licensed drivers operating an automated car in a simulator. Participants were told that due to the automation, they would not need to operate the steering wheel, brake pedal or accelerator pedal. They were instructed to monitor the roadway for vehicles that were stopped dangerously at intersections and intruding into the driver’s lane, which constituted a hazard that automated vehicles could not detect. Participants also had to distinguish between vehicles that were safely stopped and dangerously stopped at intersections.

The drivers’accuracy dropped between 7 and 21 percent over the 40-minute simulation. Even in the first 10 minutes the success rate was, at best, close to 88 percent, suggesting that all drivers missed at least some hazards.

Pat DeLucia, a professor of psychological sciences at Rice and the study’s co-author, said that one possible explanation for the results is that people get used to cars doing the driving and become complacent. Coupled with previous research that indicated people are terrible at monitoring for hazards that only happen every once in a while, and that over time their ability to respond decreases, the new study “suggests that this phenomenon of difficulty monitoring effectively over time extends to monitoring an automated car,” DeLucia said.

The bottom line is, until automated driving systems are completely reliable and can respond in all situations, the driver must stay alert and be prepared to take over. And this research clearly shows that is not happening, and gets worse as time passes.

—Eric Greenlee, assistant professor of psychological sciences at Texas Tech and lead author

The researchers hope this work will add to the growing body of research about the safety of automated cars.

These vehicles have a lot to offer, but we’re a long way from being able to detect everything going on. Until that day comes, we hope this research will raise awareness about the limitations of automated cars and their operators.

—Grenlee et al.

Resources

  • Eric T. Greenlee, Patricia R. DeLucia, David C. Newton (2018) “Driver Vigilance in Automated Vehicles: Effects of Demands on Hazard Detection Performance” Human Factors doi: 10.1177/0018720818802095

Comments

Engineer-Poet

This is my shocked face.

It's no surprise whatsoever that disengaging "drivers" from the control of the vehicle also disengages their attention.  This is normal, should be expected and is in fact the desired payoff from self-driving vehicles:  attention can be directed to more worthwhile pursuits.  What we should be doing is not trying for perfection, but making systems which are better and safer than perhaps our worst 10% of drivers.  Replacing those drivers with automation will give us a net improvement, and the data gleaned from all those vehicle-miles will yield further performance increases which will make the self-driving systems safer than the next 10% of drivers.  Lather, rinse, repeat.

Nick Lyons

@EP: Your scheme is not practical: just how do you identify the worst drivers, and then limit them to the automated cars? Testing? Special driver's license?

I believe automated driving is going to be limited to certain very well-marked, limited access highways for a long time (decades, IMHO) before we ever get to level 5 autonomy.

Engineer-Poet
just how do you identify the worst drivers...? Testing? Special driver's license?

Statistics do pretty well.  The newest and the elderly have the highest per-mile risk.

and then limit them to the automated cars?

Families often have very high insurance rates if they have new drivers.  If those drivers got lower rates with automated driver assistance, the automation could pay for itself in insurance savings.  Ditto those whose senses are going, who may have a similar per-year mishap rate but many times the per-mile rate of average drivers.  Cutting their accident rate by half might make a serious difference in both cost and quality of life.

Well-travelled roadways are going to be well-mapped by all kinds of V2x technology in the very near future.  Every tree, shrub, light pole and sign post will be mapped within inches, and differences ranking from semi-trucks to the odd wild animal will be detectable in real time and maybe in stereo or more resolution by multiple vehicles simultaneously.  Automatic control has the potential to become safer than the average driver much sooner than you think.

Lad

E. Musk says by the end of this year. We'll see.

Verify your Comment

Previewing your Comment

This is only a preview. Your comment has not yet been posted.

Working...
Your comment could not be posted. Error type:
Your comment has been posted. Post another comment

The letters and numbers you entered did not match the image. Please try again.

As a final step before posting your comment, enter the letters and numbers you see in the image below. This prevents automated programs from posting comments.

Having trouble reading this image? View an alternate.

Working...

Post a comment

Your Information

(Name is required. Email address will not be displayed with the comment.)