Driverless Car kills again

[flecc]

A driverless car being used during a trial in Arizona has killed a woman cyclist. Previously also in the USA one killed it's driver. Another has skimmed a cyclist so close that the test drive had to be aborted, and many demonstrations have resulted in a standby driver having to hurriedly grab back control. Nissan have admitted that detecting cyclists is a very big problem.

Considering how few of these have ever been tried on the roads and the whole raison d'être of these is to improve safety, I think it's high time they were taken off the roads for very long term testing in artificial situations with remote controlled dummy pedestrians, simulated bicycles and cars doing the unexpected as happens in the real world.

Against much opposition I've argued in this forum that successful driverless is a very long way off and may never happen outside of roads prepared and equipped for it. Today's trials are obviously overly ambitious and the technology is far from ready. Using humans as potential victims is wholely unacceptable.
.

 

[anotherkiwi]

I can understand the use of driver-less cars on roads without pedestrians and cyclists. The problem is getting to those roads...

 

[oyster]

whole raison d'être of these is to improve safety

I suspect you are wrong. The raison is surely to make money?

My own view is that even some relatively simple systems could make quite a difference to safety without going all the way to driverless. For example, the various assist systems, enhanced parking sensor systems (mine certainly detects pedestrians walking near my car), wakefulness detection.

The rest I largely agree. In a theoretical sense, I look forward to them trying to manage the roads round here. Hoping they won't even try in reality.

 

[flecc]

I suspect you are wrong. The raison is surely to make money?

According to both the makers and polticians and many published sources, it's to eliminate the driver error element. This has been supported by previous members comments, it's definitely safety.

Personally I think this is the opposite of technology to make money for the manufacturers. It takes the driver enjoyment out, cutting those sales, and it encourages the replacement of personal cars by far fewer mass transport fleet ones, like the driverless Uber one that has just killed. Car sales could be cut by a factor of ten or more.

My own view is that even some relatively simple systems could make quite a difference to safety without going all the way to driverless. For example, the various assist systems, enhanced parking sensor systems (mine certainly detects pedestrians walking near my car), wakefulness detection.

Definitely agreed, though whether people who can't park their car should be on the road is another question.

The rest I largely agree. In a theoretical sense, I look forward to them trying to manage the roads round here. Hoping they won't even try in reality.

Ditto. I don't think they will ever be able to handle all the road types and situations we have now. I have a mental picture of two facing each other across a one car gap and patiently waiting for ever for one to give way. They won't be able to communicate to resolve the problem since that ability could open them to hacking to deliberately create crashes.
.

 

[soundwave]

They won't be able to communicate to resolve the problem since that ability could open them to hacking to deliberately create crashes.

 

[Woosh]

The role of the driver in this incident has to be investigated.
His job is to take over the controls in cases like this.
How was he and what was he doing at the time of the accident?
Could he have taken over and caused the crash instead of the car?
let's wait for the result of the police investigation.

 

[Zlatan]

Driverless cars are another example of solving a problem that does not exist. The safest solution would be develop the safety systems to help driven cars avoid accidents.

 

[Woosh]

Driverless cars are another example of solving a problem that does not exist.

the problem exists alright.
Congested cities could do with more public transport which has to be cost effective to wean people off personal cars.
driverless electric taxis can be a good solution. By the time driverless systems get to level 5, the pent up demand would be enormous.

In this accident, I doubt that the driveless software is entirely to blame. The software would have stopped the car if it detects a moving object in front of it.
A weakness of the lidar sensing system is how to distinguish between traffic in the incoming direction and pedestrians crossing the road where there is no zebra crossing.
I think the cause is likely to be more than just the difficulty for the car to recognise pedetrians and cyclists.
I would force driverless cars to be painted bright orange with 'driverless' logo for pedestrians to be aware.

 

[anotherkiwi]

the problem exists alright.
Congested cities could do with more public transport which has to be cost effective to wean people off personal cars.
driverless electric taxis can be a good solution. By the time driverless systems get to level 5, the pent up demand would be enormous.

In this accident, I doubt that the driveless software is entirely to blame. The software would have stopped the car if it detects a moving object in front of it.
A weakness of the lidar sensing system is how to distinguish between traffic in the incoming direction and pedestrians crossing the road where there is no zebra crossing.
I think the cause is likely to be more than just the difficulty for the car to recognise pedetrians and cyclists.
I would force driverless cars to be painted bright orange with 'driverless' logo for pedestrians to be aware.

Better still they need an app that flashes a warning to all pedestrians phones along their route...

 

[flecc]

In this accident, I doubt that the driveless software is entirely to blame. The software would have stopped the car if it detects a moving object in front of it.
A weakness of the lidar sensing system is how to distinguish between traffic in the incoming direction and pedestrians crossing the road where there is no zebra crossing.
I think the cause is likely to be more than just the difficulty for the car to recognise pedetrians and cyclists.

A couple of reports say she was crossing the road and the front wheel of the bike I saw was badly buckled, so it was not a strike from the rear. Given how little there is of a bike below crank level or side on to be detected and the acknowledgements that there's great difficulty in detecting bikes, it could easily be a detection failure in the brief time of her crossing.

Even if it was driver inattention, it's still a good reason to take them off the road since that will be a big problem with drivers having little to do.
.

 

[Woosh]

Even if it was driver inattention, it's still a good reason to take them off the road since that will be a big problem with drivers having little to do.

I think the important issue here is the fact that the backup driver did not do better than the AI and detection software.
He should have seen the cyclist and understood her intention.

 

[flecc]

I think the important issue here is the fact that the backup driver did not do better than the AI and detection software.
He should have seen the cyclist and understood her intention.

Should have, but inattention among the under utilised is a well recognised problem.

It was a major problem in nuclear power stations which are computer controlled. To solve it monitoring staff are given a series of unnecessary duties to be performed at intervals, such functions as checking and recording a reading or pressing a button sequence. They have to keep to the time schedule intervals or a computer alarm goes off to alert the problem, and that keeps them alert.

Likewise tube trains have the "dead mans handle" to keep the driver physically active.

In trying to eliminate driver error, driver monitored driverless cars have invented a whole new problem.
.

 

[Woosh]

Should have, but inattention among the under utilised is a well recognised problem.

that is a problem for the AI software and sensors to solve.
I reckon in less than 30 years, the AI will be better than any human driver.
By then, cars will probably be fitted with transponders.

 

[rower]

Obviously it is dreadful that someone has been killed. But this doesn't put me off autonomous vehicles. The deceased was struck crossing the road - so while I am not jumping to a conclusion about blame, but if she'd come out from behind cover for example even an AV can't brake from e.g. 50 to 0 in 2 seconds. I would already rate an AV over a human driver, my dad's been to hospital over 5 times from irresponsible drivers and all the hits were with white vans or cars behaving irresponsibly.

 

[flecc]

that is a problem for the AI software and sensors to solve.
I reckon in less than 30 years, the AI will be better than any human driver.
By then, cars will probably be fitted with transponders.

Certainly getting a part-used driver out of it will be better.

I think there are two acceptable options in this order:

Full driver control but sensors having overriding control in sensed emergencies.

Full robotic control without any driver function when it becomes good enough.

I can't see any system needing driver monitoring being acceptable, since it incorporates the worst of both worlds, both fallible.
.

 

[Woosh]

Full robotic control without any driver function when it becomes good enough.

that's autonomous driving level 5.

We are at level 3/4 and may be 10-30 years from level 5.

Level 0: This one is pretty basic. The driver (human) controls it all: steering, brakes, throttle, power. It's what you've been doing all along.

Level 1: This driver-assistance level means that most functions are still controlled by the driver, but a specific function (like steering or accelerating) can be done automatically by the car.

Level 2: In level 2, at least one driver assistance system of "both steering and acceleration/ deceleration using information about the driving environment" is automated, like cruise control and lane-centering. It means that the "driver is disengaged from physically operating the vehicle by having his or her hands off the steering wheel AND foot off pedal at the same time," according to the SAE. The driver must still always be ready to take control of the vehicle, however.

Level 3: Drivers are still necessary in level 3 cars, but are able to completely shift "safety-critical functions" to the vehicle, under certain traffic or environmental conditions. It means that the driver is still present and will intervene if necessary, but is not required to monitor the situation in the same way it does for the previous levels.

Level 4: This is what is meant by "fully autonomous." Level 4 vehicles are "designed to perform all safety-critical driving functions and monitor roadway conditions for an entire trip." However, it's important to note that this is limited to the "operational design domain (ODD)" of the vehicle—meaning it does not cover every driving scenario.

Level 5: This refers to a fully-autonomous system that expects the vehicle's performance to equal that of a human driver, in every driving scenario—including extreme environments like dirt roads that are unlikely to be navigated by driverless vehicles in the near future.

 

[flecc]

Yes, seen those before. It's levels 2 and 3 that for me are unacceptable. Combining human and robotic fallibilities is worse than having just one.
.

 

[oyster]

I would force driverless cars to be painted bright orange with 'driverless' logo for pedestrians to be aware.

You could infer that it is OK if those with very poor or no eyesight are knocked over or killed? I think driverless vehicles have to be positively better than human-driven in order to be acceptable. Hence, it should be those human-driven ones that should be labelled - were that a sensible suggestion.

 

[flecc]

You could infer that it is OK if those with very poor or no eyesight are knocked over or killed?

Not a problem. By the time full driverless is ready, large numbers of electric cars running around silently will have wiped out all those relying on hearing when crossing roads.
.

 

[Woosh]

You could infer that it is OK if those with very poor or no eyesight are knocked over or killed? I think driverless vehicles have to be positively better than human-driven in order to be acceptable. Hence, it should be those human-driven ones that should be labelled - were that a sensible suggestion.

you wouldn't push a bike across a road when there is a car approaching would you?