Self-Driving Car Runs Over Pedestrian (March 19, 2018)

From these initial reports it looks like the pedestrian jaywalked right in front of the vehicle as it passed. Neither the self-driving car or the backup driver could have prevented the accident.

condolences on his Twitter Account. Thats classy.

None of these cars are ready for the road yet. They have no safety backup thats much higher on the foolproof scale. Accidents will happen but theres been about 5 this year alone and now a second fatality.

At the present time, I am not comfortable with having self driving cars on our roads, streets, and highways.

Human driven cars kill people too. Maybe in the end this comes down to statistics or particular environments where humans are just better.

the tests cant make a safe car . They only reveal what needs to be done in order to market these things. Just like testing anything that moves among us. (like jets or choppers).
We are gradually inserting driver assistance gizmos onto newer cars. stuff like side radars,tv cameras, blind spot monitorings and auto braking if the subject car is closing on another vehicle at some unsafe distance or speed. I just bought a new Explorere with all the electronic safety devices. It makes us feel like weve got a prtner in there with us, BUT, e still dont just accept what the car radar says (thats theres nothing oming to T -bone us). We still try to watch carefully

For me, while it's concerning (of course), it doesn't sour me to self-driving cars. Here's why.

See: https://seriousaccidents.com/legal-advice/top-causes-of-car-accidents/ They also mention High Winds; Loose Objects in Cars; and Police Car Chases. I realize that's a lawyer site and there don't seem to be any stats or studies referenced, but it still seems to be a decent list (although the order might be suspect - but just because maybe Ice should be #15 rather than #16 doesn't mean the list is without value).

A self-driving car can only by definition have the following possible causes: #5, #10, #16, #17, #19, and #22. Plus possibly high winds.

I'd say that at least some of the weather-related issues could be reduced (maybe not 100% eliminated) with sensors for ice or precipitation or even the composition of the oils and junk on the road surfaces. Fog might not be a problem (although it could be for a human-driven car which runs into it). Presumably the car would be programmed to slow down for curves so #23 is out, and obey traffic rules, so #6, #7, and #12 are out.

A self-driving car (assuming no hacking) can't be an instrument of road rage, or become more dangerous because of drinking or drugs.

Frankly, I think their best usage could be for the elderly and blind, to give people independence and the ability to do things like buy more groceries than just what they can carry on the bus.

I never thought about it, but this is an excellent point.

I was thinking of them as small busses to ferry folks to small vacations or cultural events.

rental units like avis or budget .

I think self-driving cars would be most useful to people with handicaps or drunks. The problem is it will be a long time before these people or any people can afford driverless cars. I read somewhere that by 2050 the majority of the cars will be self-driven. It's mostly a matter of economics. By 2050 the driverless car will add just $3,000 to the price of the car.

It would be handy for a drunk. He could stagger out of a tavern, hop in his car, punch in a few numbers, and fall asleep until he gets home. The problem is if he puts in the wrong numbers he could wake up in Fargo, North Dakota.

I don't think driverless cars cars will be programmed to break the law, that is they won't break the speed limit.

As far as the safety issue goes, anybody can become better drivers. Simply stop speeding, stop following too closely, stop weaving in and out, signaling when changing lanes, stop driving faster than the traffic, stop driving drunk or sleepy. We don't need to spend enormous sums of money to make us better drivers. If need be spend a few bucks and take a defensive drivers course.

Typical freedom hating serfs, one crash and it's banned. In order to make the roads safer you need more self driving cars not less, they should be going a lot faster too, with lances spears and chainsaws protruding from them.

Only then will the streets be safe.

It seems to me that people who claim that we need self-driving cars for safety are poor drivers themselves. People who drive defensively can avoid virtually all accidents.

The reason self-driving cars are potentially safe is because they don't drive drunk, they don't drive while sleepy, they don't speed, they don't follow too closely, and they don't weave in and out nor drive faster than the rest of the traffic. And they don't talk on cellphones while driving. These are driving techniques that anybody can adopt to become safe.

We don't have to spend an enormous amount of money to buy an autonomous car for safety when, for free, we can learn to drive just as safely.

From the video released, it looks like the car should have been able to detect and react to the pedestrian. Police investigators say that there were no indications that the car braked at all during the collision.

Graphic dashcam video of collision

Hmmm, software error? Maybe they should arrest the guy who wrote the subroutine. Or maybe it was a brake failure. Do these cars produce syslog files?

Defensive driving has been around for decades, yet people still don't do it.

Amazing video. The first thing that surprised me is how fast the car seemed to be going. I didn't know these thing traveled that fast.

From a more technical perspective, it seems to me that the pedestrian was in a very unfortunate place with respect to the street lights (not implying that it was the pedestrian's fault in any way). She was just behind the illuminated area as perceived by the camera, which may have fooled the software. I'm not sure how computers analyze video images to interpret real-world objects. I don't know if it takes a threshold of contrast to trigger actions or something, but it seems clear that they need to improve the algorithm for detecting objects outside of the contrast range of the headlights in conjunction with external sources (like street lights). I would also hope that they are using some type of radar or echo-locating mechanisms in front of the car as redundant checks. Then they need to correlate all that with the vehicle speed and road conditions to be sure it can stop before impact. Honestly I don't know how they code for all this stuff and guarantee processing occurs quickly enough. There are so many real-world variations to anticipate and code for.

Elon Musk, co-founder of Tesla Inc. has rejected radar for the Tesla Semi as too expensive and has instead opted for visual sensors. Tessa Semi is autonomous in the sense that in forming convoys the lead truck must be manually driven.

Economics alone will keep self-driving cars from being popular for a number of years. It's estimated that by 2035 the majority of cars will be self-driven, and will add $3,000 to the price of the car.

You can't make other people be defensive drivers, you can only do it yourself. Autonomous cars won't make other people better drivers just the autonomous car itself.

they will be sporting devices with much better "reflexes" than people can muster. Boredom, fatigue, etc can be eliminated as a reason for accidents on the part of the autonomous vehicle.

Its true that people cant be 100% responsive and aware, but an electronic driver will be monitoring rod and traffic conditions thousands of times per second.
If the vehicle hs a prime directive to never exceed speed postings and gets better and better at sensing, Im sure we can reduce accidents by a huge amount.

Yeh, this is a setback, but its the sunrise of the applied tech. Tunneling and underground mining machines are mostly autonomous in USA. As recently as 2000, they were still driven by a human. Now a human accompanies the machines as a "second pair of eyes".

Much of our land grading for projects having large tracts of land are accomplished by computer driven dozers , graders , an drags. All this is tied together by gps surveyed maps in their onboards.

Farming uses a series of autonomous plows, seeders , drills , and fertilizer applicators (and the fertilizer is precisely delivered based on needed chemistry and amounts on NPK) .Its still cheaper to use human drivers.
Is our airforce slow;y going over to drones and autonomous ROV's?.

Id love to go to the shore every weekend qnd sit in a lounge- like cockpit and read or enjoy a beverage, and let the autonomous car do the driving to the place .

I wanna go back to the Black Hills and do some fossil hunting but I hte driving that part of the country (also upstate Western NY)

mark