https://www.bloomberg.com/news/articles/2018-03-19/uber-autonomous-car-involved-in-fatal-crash-in-arizona

Initial viewing of the front camera video shows a cyclist stepping out of a dark median right into the path of the vehicle. It's unlikely any type of driver could have avoided that accident.

Now, that said, it's important to know what the car "saw". Even if it couldn't have been avoided, the car should be able to detect that a collision was imminent. Did the car detect the pedestrian at all? Did it mis-classify it? Did it misinterpret the data?

The human in the driver's seat claims he never knew there was anything there until he felt the collision.

With a completely automated system, it would be possible to operate vehicles at outrageous speeds mere inches away from each other - greatly shortening the travel times. With feedback, traffic control would respond to the flow rates and optimize them.

Traffic fatalities would be almost nonexistent. Of course, like air travel, the ones that did occur would be spectacular with a high body count.

I'll not be a passenger. I will putter around in my Red Barchetta.

This is the first known accident.

NoBSU said:

---

I will putter around in my Red Barchetta.

---

But only around your uncle's country place.

So now a car can actually kill people without a human operating it, and we want more of those.

What a country.

whitetrash said:

---

NoBSU said:

---

I will putter around in my Red Barchetta.

---

But only around your uncle's country place.

---

No one knows about it, though. He says it used to be a farm.

Better hope there are some narrows bridges. The motor law cops will chase you.

Dia del DougO said:

---

So now a car can actually kill people without a human operating it, and we want more of those.

What a country.

---

I absolutely cannot believe how many $$$$$ damages lawsuits these automakers are setting themselves up for. Boggles the mind.

OsoCoreyell said:

---

whitetrash said:

---

NoBSU said:

---

I will putter around in my Red Barchetta.

---

But only around your uncle's country place.

---

No one knows about it, though. He says it used to be a farm.

---

Is that uncle named Junior?

Quote:

---

This is where I thought self-driving cars were beheaded.

---

FTFY

Safer, more efficient, less need for parking lots. Can't happen soon enough.

quash said:

---

Safer, more efficient, less need for parking lots. Can't happen soon enough.

---

Plaintiff attorney's dream.

NoBSU said:

---

quash said:

---

Safer, more efficient, less need for parking lots. Can't happen soon enough.

---

Plaintiff attorney's dream.

---

Just the opposite: there will be far fewer accidents.

quash said:

---

NoBSU said:

---

quash said:

---

Safer, more efficient, less need for parking lots. Can't happen soon enough.

---

Plaintiff attorney's dream.

---

Just the opposite: there will be far fewer accidents.

---

So you hope.

Electric cars are a great idea and the future changes to semi truck idling are positive ideas that will challenge battery technology. But battery technology was where we needed it. Still isn't. Musk's plant isn't any gain there in new tech. I have seen new types that are in beta testing that may be an answer. But they have not been tested in volume over time. The Navistar bid testing will probably hash this out.

So you hope all the tech is where we need it. That real conditions will match test conditions.

NoBSU said:

---

quash said:

---

NoBSU said:

---

quash said:

---

Safer, more efficient, less need for parking lots. Can't happen soon enough.

---

Plaintiff attorney's dream.

---

Just the opposite: there will be far fewer accidents.

---

So you hope.

Electric cars are a great idea and the future changes to semi truck idling are positive ideas that will challenge battery technology. But battery technology was where we needed it. Still isn't. Musk's plant isn't any gain there in new tech. I have seen new types that are in beta testing that may be an answer. But they have not been tested in volume over time. The Navistar bid testing will probably hash this out.

So you hope all the tech is where we need it. That real conditions will match test conditions.

---

Getting humans out from behind the wheel is a big safety gain. PI lawyers will have to find a new bread and butter.

In situations like this, we'll just go see what went wrong. In essence, the car has a whole lot of sensors (cameras that take images, radar types that detect objects behind other objects, etc). The video is crummy, but has absolutely nothing to do with the car's ability to navigate.

The car gathers data, then sends it to a computer. The computer is just doing a huge, unending game of "If-then". "If" an object is in the roadway, "then" decide if it's a collision threat. "If" it isn't, "then" keep on trucking. "If" it is, "then" do something else.

Did the car not detect the woman (the sensors provided no data that she was even there), or did the decision tree have a hole in it (her presence was registered, but the logic did not indicate a need to do something different)?

These are situations that can be identified and corrected, forever making self-driving cars safer. The conundrum comes when I'm forced to program the logic for "If" there's a group of kindergarten kids running into the road, "then" do I choose to hit them or do I choose to send the vehicle (at presumably high speeds) into a wall or opposing traffic?

It's one thing for a human to make this decision spontaneously in the micro-second that his reactions allow him in an ugly situation. It's quite another for a logic programmer to ponder the answer, then hard code it into a machine.

Mr Tulip said:

---

In situations like this, we'll just go see what went wrong. In essence, the car has a whole lot of sensors (cameras that take images, radar types that detect objects behind other objects, etc). The video is crummy, but has absolutely nothing to do with the car's ability to navigate.

The car gathers data, then sends it to a computer. The computer is just doing a huge, unending game of "If-then". "If" an object is in the roadway, "then" decide if it's a collision threat. "If" it isn't, "then" keep on trucking. "If" it is, "then" do something else.

Did the car not detect the woman (the sensors provided no data that she was even there), or did the decision tree have a hole in it (her presence was registered, but the logic did not indicate a need to do something different)?

These are situations that can be identified and corrected, forever making self-driving cars safer. The conundrum comes when I'm forced to program the logic for "If" there's a group of kindergarten kids running into the road, "then" do I choose to hit them or do I choose to send the vehicle (at presumably high speeds) into a wall or opposing traffic?

It's one thing for a human to make this decision spontaneously in the micro-second that his reactions allow him in an ugly situation. It's quite another for a logic programmer to ponder the answer, then hard code it into a machine.

---

Isn't it amazing to what lengths the world will go to try to exempt human beings from personal responsibility?

bearlyafarmer said:

---

Mr Tulip said:

---

In situations like this, we'll just go see what went wrong. In essence, the car has a whole lot of sensors (cameras that take images, radar types that detect objects behind other objects, etc). The video is crummy, but has absolutely nothing to do with the car's ability to navigate.

The car gathers data, then sends it to a computer. The computer is just doing a huge, unending game of "If-then". "If" an object is in the roadway, "then" decide if it's a collision threat. "If" it isn't, "then" keep on trucking. "If" it is, "then" do something else.

Did the car not detect the woman (the sensors provided no data that she was even there), or did the decision tree have a hole in it (her presence was registered, but the logic did not indicate a need to do something different)?

These are situations that can be identified and corrected, forever making self-driving cars safer. The conundrum comes when I'm forced to program the logic for "If" there's a group of kindergarten kids running into the road, "then" do I choose to hit them or do I choose to send the vehicle (at presumably high speeds) into a wall or opposing traffic?

It's one thing for a human to make this decision spontaneously in the micro-second that his reactions allow him in an ugly situation. It's quite another for a logic programmer to ponder the answer, then hard code it into a machine.

---

Isn't it amazing to what lengths the world will go to try to exempt human beings from personal responsibility?

---

haha exactly. We can make the car learn from this accident and improve its ability to make different (better) decisions in the future to prevent the same accident from happening again.

But we can't make people who wear dark clothing at night and walk out in front of a car (which has its headlights one and is going the speed limit) in the absence of a crosswalk or crossing signal smarter...

https://www.mercurynews.com/2018/03/30/tesla-autopilot-was-on-during-deadly-mountain-view-crash/

BaylorBJM said:

---

First fatality in over 10,000,000 miles driven. That's incredible . Tragic story nonetheless but the numbers behind driverless vehicles are more than clear.

Automated cars are exponentially safer than having a human behind the wheel.

---

According to the National Safety Council there are 1.25 auto accident deaths (including the type here) per 100,000,000 miles driven so, in fact, they are not exponentially safer...they would be just the opposite of that.

I still think the technology is viable. Right now EVERY accident without driver input is being scrutinized. The question there is would a person have done any better avoiding the accident than the computer did.