Arizona halts...self-driving tests; supplier says Uber disabled Volvo safety system before fatality

L A Times ^

[Sequoyah101]

Im concerned about self driving cars but a lot less than i am the trucks. Just nucking futs.

[fuzzylogic]

You’re arguing against the reduction of 35,000 deaths on the road every year, mostly caused by distracted or drunk driving.

This vehicle was equipped with Lidar. It should have seen the person. This is exactly the use-case that is supposed to highlight how much better this technology is over humans.

It completely failed. For this reason Uber has a problem here. Many people involved in the technology are very concerned with this specific instance. With the technology available this should not have happened.

The real question is when this technology is mature enough to reduce deaths on the road. Will they cause fatalities? Yes. Will they reduce deaths on the road? Yes.

[piytar]

Thank you (and some other Freepers) for the clarification and information. Looks like I was wrong.

[TalonDJ]

Mochell is looking good. Did she get a makeover?

[Don W]

Actually, excessive speeding is the #1 cause of road deaths.

Distracted driving is hands down #2

THEN comes “impaired” driving.

You have noticed how the standard for “impaired” has been being forced ever downward, haven’t you?

[JustaTech]

I think the demarcation is the characteristic of the headlights. These days many headlights have a sharp cut-off when using low beam. A road near my house has a number of dips and little hills, so when meeting oncoming traffic at night it’s like getting flashed with high beams every time a newer car goes through a dip.

[JustaTech]

If the safe operation of a driverless vehicle depends on the sensors having full coverage (it does), then there should be a routine and required procedure to test them. Such as a BIT when the car is started, or a software feature that disables auto drive mode if nothing is detected by a given sensor within a defined quantity of time or distance covered.

It will be really, really dumb if it turns out this car had a faulty sensor that wasn’t detected by software.

[JustaTech]

According to USDOT, drunk driving is the largest primary factor in fatal crashes, though speeding and lack of being belted in run close behind.

The 2016 national data shows that:

Distraction-related deaths (3,450 fatalities) decreased by 2.2 percent;

Drowsy-driving deaths (803 fatalities) decreased by 3.5 percent;

Drunk-driving deaths (10,497 fatalities) increased by 1.7 per­cent;

Speeding-related deaths (10,111 fatalities) increased by 4.0 percent;

Unbelted deaths (10,428 fatalities) increased by 4.6 percent;

Motorcyclist deaths (5,286 fatalities – the largest number of motorcyclist fatalities since 2008) increased by 5.1 percent;

Pedestrian deaths (5,987 fatalities – the highest number since 1990) increased by 9.0 percent;

and Bicyclist deaths (840 fatalities – the highest number since 1991) increased by 1.3 percent.

Mandating universal use of self-driving cars would greatly reduce fatalities on the road. However..

The more prevalent they become the slower they will all have to move to keep the control grid from crashing, so there will be a greater number of people dying while waiting hours for the self-driving ambulance to slowly roll up to the scene. And there will be a large number of increased suicides and other stress-related deaths from sleep-deprived people who just can't face another 6-hour jerk-start-travel five feet-jerk-stop-repeat-endlessly commute to get to their job 12 miles away.

[Don W]

What a surprise, the Defensive Driving course I just completed in January used other data.

The amount of agitprop and misinformation in other sections of the course have been increasing every time. I take one about every 5 years or so.

[PAR35]

No, that might cut off the top. It doesn’t create a dark rectangle in the middle.

[PAR35]

Looks like the driver/attendant was looking down - texting, perhaps, instead of watching the road.

[KC_for_Freedom]

So true.

And if this car was a participant in some form of beta testing, there should be multiple tests of the features being evaluated. And the human back up should have been a trained engineer with capability to handle instant full control of the vehicle. (The driver was clearly not paying attention at this critical time.)

And, was the car tested at night on the track to verify it could detect and take appropriate action before it went on the live highway? Shame on UBER.

They will deserve everything that is thrown at them in this debacle. (Personally they deserve to be out of the automatic control business.)

[jiggyboy]

I could not believe that they said the lidar would have detected but was not tied into the decision to apply the brakes.

That's legit. Their product is a sensor and maybe some software that makes its data manageable. They're the lookout at the front of the ship. Their job is to yell "iceberg ahead starboard!", and it's somebody else's job to know what to do about it. They're not the whole system, Uber is.