Brake Failure (and an accident in a snowy day)

I think this is a great unknown about self-driving. First off, they relay heavily on lane lines/markers. Living rurally, many of our roads don't have them and the ones that do are usually in terrible shape. Now add in marker obscuring weather and snow.

I think the Rural North will be the very last places Driverless Cars catch on... if ever.

 

Uh, if the possibility of black ice, snow, or slush occurs on our local roads, when the street is clear of traffic I do a “micro-stop” where I momentarily slam on the brakes to empirically test the coefficient of friction. You could certainly program the ECU to do the same much more quickly and unobtrusively than a human could apply the brakes. In my experience , visually differentiating black ice from wet pavement , or freezing rain from normal rain, is almost impossible, which is why I rely on empirical testing.
But relevant to the ABS discussion, I had completely forgotten how dicey driving on deep gravel road surfaces , or “washboard “ rutted dirt roads was in the pre-ABS days. Gravel more than 1-2” thick feels roughly like driving in snowy slush, IIRC, but the washboard rutted dirt roads made both braking AND directional control quite difficult (as well as putting a lot of wear and tear on the suspension and stearing systems).
Empirical question: will washboard rutted dirt roads also set off the “cancel regen braking” phenomenon? If so, how long will the the regen braking be canceled if the uneven impacts of the ruts on the different wheels are basically continuous? Does anybody know this for sure? If not, I’m going to have to find a good section of washboard rutted dirt road and do some tests. (Especially with higher tire pressures, the tires and suspension would take a beating during those tests. Sigh.)

 

I totaled a car with great ABS brakes on glare ice. Nothing more dangerous. Can be the result of a little fog and a cold street.

Or a sharp curve, just started to rain after a long drive spell. The curve collects oil over the dry spell as people round it at speed. The oil seeps into the small pours in the surface. Once the water starts to hit that surface, the oil and water mix and come to the surface making it like ice. You need luck and great tires and a slow entry or you just turn into the curve and nothing happens as you slide ahead into what is coming.

The other day I was driving in the left lane on a city street. Dry. A car (hulking SUV) belatedly decided he wanted to get a quick bite at the burger joint and started to left turn into the center lane but left his nice person end hanging out at least 8 feet in my lane. Of course no turn signal. I saw the car in front of me slam on his brakes and I did too. As I was in full pedal to the floor mode I ran over some railroad tracks, unsettling the braking system. I kept my foot down hard and the ABS took hold after the tracks and I was on smooth pavement again. I could feel the ABS pulse repeatedly. No accident. One scared driver as I waited for the BMW behind me to have the same problem and slam into me. He stopped a foot short.

 

I've been doing these winter 'micro-stops' for traction testing too, since my late teens. Originally, it was a momentary full-force slam on the brakes, followed by recovery from the minor upset that sometimes occurred. Always performed on an empty road with no witnesses. While that was often easy in a very rural area, other times it was not often enough. The minor upsets provided some practice in skid recovery, and getting a feel for how the car responded.

Later, I switched to a gradual ramping up of brake force until something started to slide, or until feeling enough grip to be comfortable. After getting ABS, I went until the ABS triggered. This gradual method can be used on roads not fully empty of traffic, the brake lights also help alert and remind traffic behind to the poor traction conditions.

These tests have occasionally produced some major traction surprises, both positive and negative. The negative cases were particularly important, prompting reduced speeds well before purely visual cues would have suggested so.

 

The original DARPA Challenge self-driving courses were not on paved roads with lane markers. Mostly off-road, and unpaved country 'roads' without paint and markings. The military utility of this is great enough that it will happen. They made great progress even before I was hearing of any of these driver-assist technologies rolling out to the consumer market.

The current affordable consumer systems that depend on lane markers are just a first generation, applying easy short-cuts and the cheapest available sensors to cover many but not nearly all situations. As technology improves and increased volume brings hardware costs come down, I'm expecting major performance improvements. eventually covering all public roads in nearly all weather.

But we also have set up a race condition between improving semi- & fully-autonomous operation, and customer expectations. Drivers are already mentally checking out of their driving responsibilities faster than their cars can safely take over, resulting in increased traffic fatalities since 2013.

 

My perception is that the oil-on-pavement problem is less with modern cars as they seem to leak less oil. But boy those oil slicks were dangerous right when it started to rain and the surface oil emulsified into a slick film. I drove motorcycles and having either front or rear wheel sliding out from under you when you hit those oily patches was very unsettling.

 

The first uneven wheel-speed signal that fools the ECU into thinking traction's been lost (or, in some cases, correctly informs the ECU that traction's been lost) will trigger the one-time shift from regen to friction for the rest of that use of the brakes. It doesn't flip/flop back and forth.

Your next use of the brakes will start in the normal way again, and will also go to friction if the wheel sensors again trigger it to do so.

If you are making an urgent stop (as seen by the pedal stroke sensor in how hurriedly you press the pedal), regen is skipped and friction is used from the get-go, so there's no chance of a regen-to-friction transition in that case.

-Chap

 

I would hope that it would be pretty easy.
IF.....the outside temperature is 35F or less, the car won't function in "driverless" mode at all.

I think this whole thing is a BIG bag of worms.
If the "driverless" function works 90% of the time or more, then over a period of time most of the drivers won't KNOW how to drive at all when it becomes necessary for them to take over.

Of course, a good case can be made that a LOT of people don't know how to drive NOW.

 

Curious you should say that, and I do believe that you are correct.
In aviation, an over-reliance on cockpit automation in the short term led to pilots losing some degree of situational awareness of the flight environment, and in the long term led to an erosion of “stick and rudder” flight skills. At least one airline decided to eschew automation use , and at least one airline suggested periodically turning off the automation and flying manually to preserve critical flight skills.
Second odd piece of evidence is that military pilots with many automated flight and weapons systems were found to turn off the automated systems as they were about to engage in aerial combat. I did not get to interview them, but my opinion is that all pilots HATE to be surprised by unexpected automation-caused flight inputs. Thus to avoid surprises, I would also turn off automation.
Transitioning back to the automobile domain, i suspect similar automation-related human failures will occur. Has not Tesla already had some crashes by drivers’ over-reliance upon their “autopilot” system?
(Personally, if an automated backup system such as the ABS braking circuit has to engage in an unplanned manner while I am driving, I consider that to be a “fail” and retrain myself. So far in 5 years and 80,000 miles, the ABS system kicked on unexpectedly exactly once and I had indeed failed.)

 

I told everyone that I tested MYSELF, but not published, hence no citations. Unlike many posters, I don't make up things when I post.

thanks for finding the published data. This mirrors my exact experience in snow. I still like my ABS though.