Self-Driving Cars -- Needs of the Many vs Numero Uno

[ERD50]

Quote:

Originally Posted by Texas Proud I would have told the instructor to screw it... I do not care what color the car behind is, I just care IF there is a car and what it is doing.... Even today I do not look long enough, nor is my brain programmed to process the color... the only time I actually look is when stopped because my DW has this game called 'punch buggy' and I want a leg up when I can actually pay attention....

Well, perhaps the correct answer was "No, there's no car behind me"? I thought it was a good thing for him to do, it made the point to me (a dumb teenager) about always being aware of where cars are around you, and the lesson stuck with me (as a dumb adult).

-ERD50

[NW-Bound]

Quote:

Originally Posted by NW-Bound People who do not heed the above warnings must not value their life highly, nor that of innocent bystanders.

Quote:

Originally Posted by HFWR Just like most drivers of non-autonomous vehicles...

Careless people will always find ways to harm themselves, and others.

The point I have been trying to make, in vain it seems like, is that the public has been so excited about this so-called "autopilot" that will allow them to go to sleep, or watch DVD. Or it will drive them to the hospital when they are infirm or invalid.

I surely hope I will live to see that happens. But the current technology ain't there yet. Stop dreaming. When the "real" stuff happens, we will know.

In the mean time, people who perpetuate the myth of the current autopilot are doing harm to the public. Lots of people do not know the limitations of current technology, and they also do not read manuals. They prefer to see youtube videos of "Look Ma, no hands".

[NW-Bound]

A while back, I asked the following question:

Quote:

If the computer in your car fails, is it going to accelerate hard, or stomp on the brakes? Does it swerve hard left or right? Keep coasting straight at the same speed? Or is it going to be different and unpredictable for every failure? I guess eventually the car will stop when encountering sufficient obstacles.

It was not strictly the computer that failed in the Florida fatal Tesla accident, but the camera that was supposedly blinded by the bright sky (according to Tesla).

I believe the camera was mounted inside and high up (by the rearview mirror?), and the top of the car was scraped off by the underside of the trailer. The owner of the house where the car stopped said that its top was peeled back like the lid of a can of sardines. Of course, the camera was obliterated, which means the "autopilot" was now blinded.

Still, the car kept on careening down the road, veered off and went through two fences until it was stopped by an electric pole. See photo below.

So, I guess I have my answer, which is really obvious. All cars eventually stop when encountering enough obstacles.

[JoeWras]

Quote:

Originally Posted by NW-Bound The point I have been trying to make, in vain it seems like, is that the public has been so excited about this so-called "autopilot" that will allow them to go to sleep, or watch DVD. Or it will drive them to the hospital when they are infirm or invalid.

NW: relax. It is not in vain. I like ER forums, but a lot of time people just like to talk. All these "introverts" in real life really like to spout off here, sometimes just babble. They may be listening but still need to just talk.

Now, some will disagree with your point and do feel that since self driving cars are going to be a clear revolution in transportation, they are "wish casting" it along. This is normal. See space flight as an example.

What I find interesting in this discussion is that most of us asking for patience are engineers who have seen the reality of developing products. Conversely, some of my engineer friends have even bigger blinders on, and worship at the alter of Elon Musk. It is strange.

One of my upper level managers called us into a meeting. During the meeting, the guy got off track and spent nearly 1/2 hour evangelizing about his Tesla, and how he was excited to get another software update. Frankly, it bordered on creepy.

I'm a technologist, but I don't like it when people worship technologists.

[NW-Bound]

The followers worship Musk because he dares go where others did not. That's fine, as long as nobody has to lose life or limbs, particularly bystanders.

Not being into auto, I never follow news about any car, Tesla included. But just since last week, I started to get more curious about exactly what sensors the car has to allow it to be an "autopilot". Here's what I find out.

It has a front long-range radar, mounted in the nose grill. It has a front-looking camera, most likely mounted high behind the windshield, probably on the rearview mirror. Around the car perimeter are several ultrasonic sensors. These serve as proximity sensors, and supposedly good to 16 ft. The latter are the ones that fail to pick up an adjacent high truck, as noted by a Tesla owner in an earlier post in this thread.

So, if the front-looking camera failed to detect the white semitrailer against the sky in the fatal accident, then why did the front radar not help? This has been bothering me.

I found the answer. Musk himself in a tweet said that the radar picked up the semi, but this reading could not be used because high overhead traffic signs would have the same radar signature.

So, in order to declare that there's an obstacle ahead, both the camera and radar have to agree. Else, there would be a lot of false positives. The car would brake all the time on the road, and get rear-ended.

One should now understand why the driver's eyes must always be on the road to be the tie-breaker. And he must act fast enough to save himself.

By the way, MobilEye, the company that supplies Tesla with the camera vision software says that its system does not yet deal with cars crossing the road, but its new software will in 2018. They call this "Lateral Turn Across Path (LTAP) Detection".

That's what I have found so far. Much of this hot self-driving car software is too new and proprietary, so it is difficult to know exactly what it can do. However, the above info is from the horse mouths.

[Which Roger]

As the technology advances, self-driving vehicles should be able to coordinate with each other. In the case of the Tesla accident, a self-driving semi could have detected that the Tesla was approaching at high speed, and then either taken evasive action, or communicate back to the Tesla that it was about to get in trouble. Even if some vehicles are operated manually, they could be "smart" vehicles that provide that sort of feedback to other vehicles.

[NW-Bound]

A truly autonomous vehicle is more desirable. A tree may fall across the road, or a load of brick might have fallen off a truck. A flash flood has washed out half the road. A human driver would have no problems stopping for them. We would want a car with no steering wheel and no brake to do no less. Anything less is just "driver assistance".

[HFWR]

Quote:

Originally Posted by NW-Bound Careless people will always find ways to harm themselves, and others. The point I have been trying to make, in vain it seems like, is that the public has been so excited about this so-called "autopilot" that will allow them to go to sleep, or watch DVD. Or it will drive them to the hospital when they are infirm or invalid. I surely hope I will live to see that happens. But the current technology ain't there yet. Stop dreaming. When the "real" stuff happens, we will know. In the mean time, people who perpetuate the myth of the current autopilot are doing harm to the public. Lots of people do not know the limitations of current technology, and they also do not read manuals. They prefer to see youtube videos of "Look Ma, no hands".

Apparently my quips do not always evoke the expected response...

I think "self-driving" autos would perform well in situations where the infrastructure is designed for said autos, say an interstate highway, with known access points, well-marked lanes, the capability for autos to communicate with each other, etc.

Dealing with texting, sleeping, drunken, careless, stupid, or otherwise occupied humans, on busy streets with many obstacles, is, as you've mentioned, another matter.

[NW-Bound]

Well, if something works that well as everybody including myself wants, why not?

I have often said I stopped being a car enthusiast somewhere in my 30s, a geezer before my time. And I do not care to drive. But that does not mean I will endorse or use something that has the potential to give me a heart attack when it hiccups.

See my post right above talking about what sensors are in the Tesla, and their limitations.

[Midpack]

Could it be more instructive to research how fully autonomous cars like Google's function, instead of focusing solely on the Tesla Model S? The Models S has 'autopilot' features that are only intended to assist a fully engaged driver. It's clear with this controversial Model S fatality, the driver was expecting more of the 'autopilot' feature than Tesla intended. While Tesla is working toward fully autonomous, as are many automakers, I don't get the sense Tesla is leading the way technically (even though they seem to be marketing/promoting more than others).

[JoeWras]

Quote:

Originally Posted by Midpack Could it be more instructive to research how fully autonomous cars like Google's function, instead of focusing solely on the Tesla Model S?

Well Midpack, I think they are both instructive because they take different points of view.

Your timing on this question is good. The LATimes examines that exact comparison today:
Tesla and Google are both driving toward autonomous vehicles. Which company is taking the better route? - LA Times

Quote:

“Having developed software and hardware products … I can point to the incredible inventiveness of customers in doing things you just never, ever considered possible, even when you tried to take the ridiculous and stupid into account,” said Paul Reynolds, a former vice president of engineering at wireless charging technology developer Ubeam. “If customer education is the only thing stopping your product from being dangerous in normal use, then your real problem is a company without proper consideration for safety.”

[JoeWras]

More from the LA Times article.

Quote:

The NHTSA ranks self-driving cars based on the level they cede to the vehicle, with 1 being the lowest and 5 the highest.Tesla’s autopilot feature is classified as level 2, which means it is capable of staying in the center of a lane, changing lanes and adjusting speed according to traffic. Google is aiming for levels 4 and 5 — the former requires a driver to input navigation instructions, but relinquishes all other control to the vehicle, while level 5 autonomy does not involve a driver at all.

I didn't know the NHTSA already has a scale in mind. This is good, because I think there is a disconnect in our discussions based on what a self driving car is. I see "self driving cars" as a "level 5" activity only. Others see the dream being real at levels 3 or 4.

[Totoro]

Wonder what self-driving cars will be called?

Horseless carriage became auto-mobile, and then car & auto.

Probably still cars. Maybe scars? (self-cars? sar?).

[NW-Bound]

Quote:

Originally Posted by Midpack Could it be more instructive to research how fully autonomous cars like Google's function, instead of focusing solely on the Tesla Model S? The Models S has 'autopilot' features that are only intended to assist a fully engaged driver. It's clear with this controversial Model S fatality, the driver was expecting more of the 'autopilot' feature than Tesla intended. While Tesla is working toward fully autonomous, as are many automakers, I don't get the sense Tesla is leading the way technically (even though they seem to be marketing/promoting more than others).

Thank you!

I have been trying to compare Google's approach throughout the thread, and that just did not catch people's attention (that's why politicians know to keep repeating something about 100 times before people get their message).

The LA article is good in comparing the two companies' philosophies. But let me add a bit of technical info for the curious.

Basically, Google has a far superior sensor in the Lidar. Being a laser beam, it can pinpoint distances to obstacles along very narrow lines of sight, unlike a radar beam which would need a rotating dish like the aircraft tracking ones for precise tracking (not something like the Tesla's radar that cannot tell a semi from an overhead highway sign).

They then use the vision camera to look at the objects and try to identify them. They try to track all objects around the car, and for moving vehicles try to see if their paths would intersect with their own. They would not miss the semi-trailer crossing in front of them.

So, why is Google not yet releasing their software? It's because they want to build a true autonomous car, one that needs no steering wheel. For that, the job becomes much much tougher, as they need to watch out for everything that can go wrong. They need to drive around obtacles, read hand gestures of policemen and construction workers directing traffic. They need to read street signs, temporary construction signs. Can their lidar detect pot holes for example? I saw some examples of the lidar scan, and I was still dubious because the step curbs of perhaps 5" were barely visible.

They know they are not ready. When will they be? They say soon, but I am still a bit skeptical. Their software is very impressive though.

By the way, as I repeatedly pointed out, the current lidar is an expensive dome ($75K initially) that sits on top of the car. Tesla owners who want a sleek looking car would not care for it.

PS. When I was still working, there were companies looking to use laser beams in helicopters for wire detection (a chopper rotor getting tangled in wires or power lines is very bad news). They reported that the laser worked well to detect old and dull wires which scattered the laser back to the emitter. A new shiny wire would reflect the beam like a mirror (not scattering), and the reflected beam would not hit the receptor. So, shiny objects are invisible to the lidar.

I do not know how this problem is solved with current lidars. What I told happened 25 years ago, but it's hard to cheat the laws of physics.

[Totoro]

Quote:

Originally Posted by NW-Bound I do not know how this problem is solved with current lidars. What I told happened 25 years ago, but it's hard to cheat the laws of physics.

I think Google's car also has a short-range radar. That might help?

[NW-Bound]

Probably, but only for detection of large objects. Or having multiple vision cameras to use stereoscopic effects would be a lot better than being a cyclop. Pot holes are tough though.

There are a lot of decisions the computer must make if it has no human to help. It would have to know not to drive on a soft shoulder after a rain and gets stuck. It has to detect and avoid puddles or slow down to avoid hydroplaning or splashing pedestrians and other cars... It's a lot more involved than detecting a semi-trailer. One can go on and on.

[aja8888]

Quote:

Originally Posted by NW-Bound

I wouldn't be caught dead in something like this that looks like an oversize marshmallow.

[NW-Bound]

So, you don't think the cool Lidar on top makes up for the vehicle shape?

Hey, if it is affordable and reliable, I will be the one to have it. However, I would want it larger and beefier to protect me from the texting idiots who will still be around. For me, function and safety come before form.

I will even pay a lot more if it comes with an automatic turreted machine gun that's computer controlled.

PS. By the way, the vehicle shape and the high position of the Lidar allows the latter to look down very close to the vehicle. I believe that's intended. Far superior to the ultrasonic proximity sensors in other production cars. Sitting up high, that Lidar has 360-degree vision. Excellent and lots of data for the computer to analyze.

[aja8888]

Quote:

Originally Posted by NW-Bound So, you don't think the cool Lidar on top makes up for the vehicle shape? Hey, if it is affordable and reliable, I will be the one to have it. However, I would want it larger and beefier to protect me from the texting idiots who will still be around. For me, function and safety come before form. I will even pay a lot more if it comes with an automatic turreted machine gun that's computer controlled. PS. By the way, the vehicle shape and the high position of the Lidar allows the latter to look down very close to the vehicle. I believe that's intended. Far superior to the ultrasonic proximity sensors in other production cars. Sitting up high, that Lidar has 360-degree vision. Excellent and lots of data for the computer to analyze.

Whatever the self driving car manufacturers come up with, they would still have to meet the NHTSA safety standards for bumpers, air bags, glass type, etc, etc.

I realize that little marshmallow is a test vehicle, but it is butt ugly.

[Music Lover]

Quote:

Originally Posted by aja8888 I wouldn't be caught dead in something like this that looks like an oversize marshmallow.

Does it come with a remote control??