I Brake for Autonomous Vehicle Braking

By Ernest Worthman, AWT Exec. Editor, IEEE Sr. Member

October 17, 2019

Ern’s Perspective

Worthman

Trust me, I have no intention of trusting autonomous vehicle braking.

One of the terms we see pop up in almost every technical vector is autonomous vehicles. As with 5G, the autonomous vehicle landscape is fraught with hype. That has even spilled over to the consumer marketing arena with tons of ads for automobiles showing hands-off braking, lane navigation, self-parking, and more.

Depending upon with whom one speaks, autonomous vehicles are anywhere from level 3 to level 5. Of course, the only one who believes we are at level 5 is Elon Musk, with his claims for Teslas.

Before I go much further, remember, I believe that level 5 will not be achieved until there is advanced V2 everything – vehicles, infrastructure, pedestrians, the cloud, and more. And, here is one example of why.

The AAA recently released a report on the effectiveness of Automatic Emergency Braking (AEB). If you believe the TV car manufacturers’ ads, this is the answer to not hitting a child running out into the street between two parked cars.

If you believe the AAA, that is only the case 11 percent of the time. 89 percent of the time, in the experiments conducted by the AAA, the car hit the child.

There are other not-so-impressive statistics as well. In other tests, the vehicle:

·      Struck two adult targets standing on the side of the road 80 percent of the time
·      Hit an adult target who was right around the corner 100 percent of the time
·      Were generally ineffective at higher speeds.
·      Could not detect a pedestrian at night

This is not good news. If you listen to the hype, AEB has nailed it. If you look at the AAA statistics, AEB is not even close to ready for prime time. It may work if one is about to back into a trash can in broad daylight, but the gray areas, where it is needed the most, AEB fails miserably.

Let us unpack this a bit and look into why this technology is so ineffective.

One of the issues is speed. Whether it is the driver controlling the brakes or the car has AEB managing them, if one is traveling at 30 mph and someone darts out in front of the vehicle, the stopping distance is the same, under identical conditions. There are some efficiency gains with AEB (reaction time is much better and brake application is optimized) but the stopping distance is not diminished by that much from an aware driver. Therefore, if the target is inside of the stopping distance by enough of a margin, AEB will not do a better job than a driver.

However, there are some circumstances where AEB is better than the driver. One is distracted driving. Another is impaired visibility (driving into the sun). In such cases, the “mechanical” eyes of the sensors can improve braking, markedly.

Another condition is “like” differentiation. By that, I mean the ability to differentiate between, or among, similar objects. One of those cases is at dusk. AEB has shown to fail when an object that has similar color characteristics as their surroundings. This is particularly concerning with animals such as deer when they dart onto the roadway. Another condition is a white car against a snowbank.

Interestingly, the issue is not the AEB per se, but the sensors that control it. AEB is a go/no go condition. What triggers it is the challenge.

Most of the sensors that link to AEB are visual – cameras. So AEB is only as good as the data feeding it. There have been significant improvements in camera technology in the last year or so. However, it is still a challenge for cameras, to work at the edges, even if there are multiple cameras and redundancy is implemented.

This circles back around to V2X. With radio communications, reliability can be improved by an order of magnitude, at least. Looking far into the future, let us say all clothing has some sort of transmitting capability. Micro transmitters in the heel of a shoe, pants with buttons or snaps that are micro transmitters. I could go on here forever, but one gets the point.

With such two-way communications ability, the child darting out from between parked cars can be identified long before the visual contact. The same for individuals crossing the street at a blind intersection. There are countless other use cases but, again, one gets the idea.

The bottom line here is that the hype is creating a false sense of security. If one believes it, we are only a hop, skip and a jump away from being relieved of the drudgery of driving. However, the AAA report shows we are miles away from that. And AEB is just one of hundreds to thousands of systems in an autonomous vehicle. Each has its own issues and each has its state along the reliability path.

However, this is not a “close enough is good enough” situation. This is life safety, and giving the driver a false sense of security by leading them to believe such applications are foolproof is irresponsible.

In life safety, is five-nines even good enough?