GM studying operator behavior in semi-autonomous vehicle operation; increased driver focus on road ahead
20 June 2012
General Motors and its research partners recently studied how non-driving activities influence driver behavior in self-steering, semi-autonomous vehicles. The researchers are preparing a report on the work, which is still a few months away from publication; however, one key finding is that driver attentiveness can be improved through advanced driver assistance and safety features.
The GM study examined the demands on the driver’s visual attention in hands-on steering and automated steering, both with full-speed range adaptive cruise control engaged. The studies took place in a driving simulator at Indiana University-Purdue University in Indianapolis and with Virginia Tech Transportation Institute (VTTI) on a GM test track in Michigan.
| ”Drivers are already engaging in risky behavior, and are likely to continue doing so given the prevalence of smartphones and other portable electronics, so why not make it safer for them and the people around them. Offering some form of vehicle automation with the proper safeguards might be better than what is happening on our roads today.” —Dr. Eddy Llaneras, VTTI |
More than one in three drivers surveyed in the AAA Foundation’s 2011 Traffic Safety Culture Index admitted to texting or emailing while driving, despite the vast majority of those surveyed agreeing such behind-the-wheel activities are unacceptable. When engaging in non-driving activities, drivers tend to split their visual attention between the roadway and secondary tasks by making relatively frequent, but brief off-road glances.
The study showed that advanced driver monitoring and assistance features, such as Forward Collision Alert, increases drivers’ focus on the road ahead by 126% when automated steering is in operation, which increases detection and response to roadway events.
People have dreamed of having self-driving cars for decades, but having that capability will be a major adjustment for people when it is first introduced. This study is helping GM and its research partners determine the best methods for keeping drivers engaged.
—John Capp, GM director of Global Active Safety Electronics and Innovation
The foundation for these future systems is the Driver Assist Package (earlier post) that will be available in November on the new 2013 Cadillac XTS and ATS sedans. The package includes features such as full-speed range adaptive cruise control and automatic emergency braking, which are designed to help prevent collisions caused by human error. The human factors research underway is helping GM and its suppliers identify what new technologies will be needed to ensure safe operation of future autonomous systems.
When asked, some study participants expressed strong interest in having a vehicle that could drive itself, particularly for long trips when lane centering and full-speed range adaptive cruise control could help lighten the driver’s workload.
The Federal Highway Administration helped pay for the study but does not necessarily endorse all its findings.
GM, its partners and others are continuing to study the effectiveness of the driver monitoring systems, new ways drivers interact with their vehicles, and other active safety technologies that are key enablers for autonomous vehicles. These technologies are expected to ease drivers’ workload when traffic and road conditions allow, but they are not intended for drivers to completely “tune out.”
We recognize that autonomous vehicles will require robust safeguards. By studying driver behavior in automated driving scenarios we are better able to identify the types of driver assistance and safety features that automated cars will need.
—John Capp
GM and VTTI have collaborated on a range of projects over the years, and last year completed work for the “Backing Crash Countermeasures” project, part of the US Department of Transportation’s Advanced Crash Avoidance Technologies (ACAT) program. As part of this project, they developed a basic methodological framework and computer‐based simulation model to estimate the effectiveness and potential safety benefits of various backing crash countermeasure systems.
With more more vehicles on the roads and more driver distractions more electronic assistance will be required to:
1. keep drivers more alert.
2. correct bad driving behaviors.
3. avoid childish risky driving.
4. block drunken/drugged drivers.
5. automatically adjust and maintain speed limits.
6. block risky lane changing.
7. keep safe distances between vehicles.
As about 80% of all fatal car accidents are caused by inadequate driving skill and/or practices, built in on-going driving skill/practices test could be used to restrict drivers licenses.
Posted by: HarveyD | 20 June 2012 at 01:59 PM
Great! Gidgety gadgets for morons who can't control their need to be connected to the childish endeavours. I think it's time to stop worrying about the safety of these permanently immature individuals. If you can't grow up and pay attention, perhaps a filtering process is called for, let's call it "natural selection".
Posted by: Brotherkenny4 | 20 June 2012 at 02:04 PM
" . participants expressed strong interest . . particularly for long trips when . . could help lighten the driver’s workload."
That's when the workload is really huge, yup, yup.
They say;
"We recognize that autonomous vehicles will require robust safeguards."
- Yes, like making them illegal.
Posted by: ToppaTom | 20 June 2012 at 06:57 PM
B4...unfortunately those 'morons' or 'drivers with frequent moron/childish behavior' constitute about 40% of all drivers on USA's roads, specially on Friday nights and week ends. If you could find ways to convince them to stay home or walk or use taxis you could reduce fatal accidents by 80% or by about 26,000+ every year and serious injuries by 5 to 8 times more.
USA roads create more injuries and fatal accidents than the last 4 wars in the same time frames. Unfortunately, those 'morons' kill and/or injure many other people (over 200,000 a year) that had nothing to do with those cases.
Any gadgets to correct that is welcomed.
Posted by: HarveyD | 21 June 2012 at 07:57 AM
Never mind workload; it is not intuitively obvious that this will improve driver attentiveness, regardless of what they saw in the driving simulator.
It IS obvious to us (and GM, I trust) that safety is the overriding, overarching, prime, most important, #1 requirement.
Gadgets to sell cars are fine, but I am sure they realize that one death due to a semi-autonomous GM(onster) will make them regret the whole thing.
Posted by: ToppaTom | 21 June 2012 at 12:11 PM
Seriously, ToppaTom? Human drivers kill a million people a year worldwide, and *one death* from an autonomous vehicle will make them "regret the whole thing?"
Typical cognitive bias against new technology: exaggerate the potential dangers beyond all reason, while minimizing the insane carnage of the status quo. Nicely done.
Not only are autonomous vehicles coming. At a certain point, they will be mandatory on most if not all public roads.
Posted by: Sean Prophet | 21 June 2012 at 04:30 PM
Smart radar/cameras can do an excellent job to detect and automatically penalize (stiff fines) drivers with moron like driving attitudes. Secondly, it can become a major source of needed revenues to repair roads-streets and bridges. One $15K unit installed in a good location generated $18,000,000 in revenues last year and reduced accidents/incidents by 75% in the area, probably saving another $18,000,000 or more in physical and health care cost.
Why not install 1,000,000+ of those radar/camera gadgets in good old USA$? It could save 20,000 + lives a year and many $$B in secondary cost. It could make USA roads save again to use.
Posted by: HarveyD | 21 June 2012 at 07:36 PM
"If you can't grow up and pay attention, perhaps a filtering process is called for, let's call it "natural selection"."
A few years ago I was hit head-on by a drunk driver. I was grown up and paying attention. The drunk came around a curve at a very high speed and on my side of a narrow road. I had no way to escape.
I spend a week in intensive care. I was the one who almost got selected out. That's the second time I was seriously injured by a drunk driver.
I'm all for equipping vehicles with a bit of intelligence so that they can self-avoid human mistakes.
A car seriously over the speed limit and out of its lane in a no-pass zone or a car failing to decelerate as it approaches a stop sign? Let a computer take over, obviously the driver is screwing up.
(Full disclosure: I fell asleep once while driving and ended up in a corn field. I wouldn't have objected at all if my car had either awakened me or stopped on its own. I could have not only selected myself out, but also my passengers and possibly people in other vehicles had I veered left rather than right.)
Posted by: Bob Wallace | 22 June 2012 at 10:34 AM
We have a lot of deer killed on roads around where I live. And occasionally we loose a driver as well.
What often happens is that a deer will suddenly cross the road in front of an oncoming car. Drivers generally don't hit that deer, they hit the second deer that crosses behind the first. Two or three deer crossing in sequence is common. I suspect the drivers are watching the first deer and fail to see the second in time.
Computers with lots of sensors can watch all sorts of things at once. They can even be programmed to look for "that second deer", something that many drivers don't know about.
"Oh, computer detects ball rolling across street. Set alerts for child chasing behind." (Driver watching ball to see if oncoming cars hit it....)
Posted by: Bob Wallace | 22 June 2012 at 10:41 AM
BW...I fully agree with you that smart future electronic driver assistance could drastically reduce the number of serious car accidents and related casualties. Coupled with road side automated radar/camera units, many faulty drivers could be detected, fined and/or removed from roads and streets.
It is surprising that so little has been done to stop that ongoing calamity and that so many seems to object to deal with it.
Posted by: HarveyD | 22 June 2012 at 01:44 PM
Alert/safety systems should be the goal, not "self-steering, semi-autonomous vehicles".
It should be obvious that safety should be the #1 requirement, not new technology gadgetry.
Posted by: ToppaTom | 22 June 2012 at 03:00 PM
TT. The average undisciplined driver (young and not so young) will ignore on-board alerts (visual/audible) or would quickly try to bypass them.
If we are serious about safety and security on our roads and streets we need much more than on-board alerts.
1. mandatory professional extensive training for all first license drivers and all those who lost their 8 or 10 demerit points.
2. Improved ways to detect bad driving behaviors 24/7 in as many places as required is a must. (The $15K automated radar/camera mobile/fixed stations have a very short pay back period (a few days) if location is well done.)
3. Since bad driving detectors cannot be installed every where in the world, on-board bad driving detection is also required.
Posted by: HarveyD | 23 June 2012 at 07:27 AM
"Alert/safety systems should be the goal, not "self-steering, semi-autonomous vehicles"."
Alert systems should be the first goal. I agree with that. Alert systems are the simplest to build and would provide a proving ground for future developments.
(BTW, when I was 11 I was hit from behind by a car. The driver was looking at spectacular spring flowers on the other side of the road and wandered out of the lane. An alert system could have saved me a concussion and a lot of road burn.)
But I want to see us move on to fully automated driving. There's nothing quite so aggravating as spending time in stop and go traffic. Why not let the car move one car length at a time and stop on its own? That would be a first step in a self-driving system. Limit it to speeds under 20MPG.
Then, if you've done a long stretch on an interstate. I'd love to turn that over to a computer. I already lock the cruise control down so that I don't start following someone who is speeding and end getting a ticket. I'd rather watch the scenery or whatever.
Build the systems in a little at a time. We've already got cruise control. Some cars now have automatic parking. Some have blind spot alerts. Expand function as each phase is perfected.
I wouldn't mind phoning my car and having it pick me up in front of a restaurant and drive me home. Then I could enjoy that second adult beverage.
Posted by: Bob Wallace | 23 June 2012 at 08:13 AM
Yes BW...it is going to be (de facto) progressive. However, many of the technologies required already exist and could be incorporate (as options) starting with the most useful one, such as:
1. blind spots coverage and alert.
2. wandering off the lane alert.
3. speeding alert + automatic slow down
4. driving to close alert + automatic slow down.
5. changing lane too often alert.
6. changing speed too often alert.
7. over acceleration/deceleration alert.
8. zig-zag alert.
9. recording, computing and displaying all alerts created.
10. alerting drivers to go for re-training.
Unfortunately, all those alerts may or will have most positive effects on the 60% of existing good drivers. The other 40% will not buy the options or ignore all alerts. Effective road/street side detectors (automated radar/camera) may be the only way to positively connect with them.
By the way, the neighboring police stations are alerted as soon as a bad driver, who has not paid the charges for previous violations, is detected by one of the automated radar/camera station and another violation is automatically issued. RED alerts are initiated for his/her arrest (and car seized) after a set number of unpaid violations.
Posted by: HarveyD | 23 June 2012 at 09:12 AM
"The other 40% will not buy the options or ignore all alerts."
Cars are starting to have "black box" memory. A driver who gets in an accident and for whom there is data showing that they ignored warnings would be looking at bigger trouble.
It's quite likely that it's not going to make financial sense to build cars without the systems installed. Like manual crank windows and manual transmission, it would be cheap on a one:one basis but adding changes to the assembly line and more model selection costs money.
--
I agree that we could stop some bad driver with roadside detection, but that is likely to be a harder sell to the general public.
I do expect that we'll see "intelligent" highway cameras soon. We've got quite affordable cameras with face and smile detection, detecting a car wandering out of its lane wouldn't be a stretch. Triggering a call to a patrol car to pull it over and check things out would be easy. This can all be computer-enabled notification.
That way we could avoid the problem of automated tickets. Drunks would be caught by the stop. Overly aggressive drivers would suffer the inconvenience of sitting on the shoulder for a while during their processing time.
Posted by: Bob Wallace | 25 June 2012 at 11:58 AM