Slashdot Mirror
Self-Driving Car Speed Race Ends With A Crash (electrek.co)
An anonymous reader writes:On a professional track in Buenos Aires, fans watched the first Formula E auto race with self-driving electric cars. "Roborace's two test vehicles battled it out on the circuit at a reasonably quick 115MPH," reports Engadget, "but one of the cars crashed after it took a turn too aggressively. The racing league was quick to tout the safety advantages of crashing autonomous cars ('no drivers were harmed'), but it's clear that the tech is still rough around the edges." Electrek is reporting that the cars "still have a cabin for a driver but neither car's cabin was occupied during the event." The ultimate goal is to have several teams racing the exact same self-driving car, while letting each team customize its car's driving software.
An Argentinian journalist shared footage of the race cars on Twitter, and apparently at one point a dog wandered out in front of an oncoming race car. But the real question is how the fans are going to feel about watching a speed race between cars with no drivers?
An Argentinian journalist shared footage of the race cars on Twitter, and apparently at one point a dog wandered out in front of an oncoming race car. But the real question is how the fans are going to feel about watching a speed race between cars with no drivers?
No problem, they can replace the audience with robots.
I have written a truly remarkable program which this sig is too small to contain.
The physical risk to the driver, and the driver's skill under pressure are what makes watching motor racing exciting.
Take them both away by replacing it with software and all you have is another boring nerdfest.
No excitement means no spectators. No spectators means no money. No money means no sport.
People watch racing because there is risk of a crash with humans in the cockpit. People drive in professional racing because there is a risk to themselves. Those things translate into money, jobs, technological advancements in vehicles (performance and safety). Take away the human element and it's like sitting and watching airplanes fly. Interesting for a few visits, but no sustainable market and not really entertaining. Put up a bar and bleacher stand, and it would be mostly empty.
Hell, look at the robot warrior events, which are cool but don't make money for any duration of time.
If they are doing this to build safety, no spectators needed. IMHO, bit whoop. Sarcastiball anyone?
-The wise argue that there are few absolutes, the fool argues that there are no probabilities.
I think in the early days, these races might be entertaining.
I can imagine that eventually some kind of optimum strategy may evolve and all the teams use it, and then the cars will all do the same thing and the race will be boring. But in the early days, with people trying different strategies, stuff might happen that is interesting to watch.
I remember back at my first job, we found some kind of game where you wrote a program to control a robot tank in the game, and the whole game was to have matches between people's programs. The programming language was simple and there were APIs for things like "throw out a radar ping", "turn tank", "rotate turret", "fire gun", "check to see if tank is damaged", etc. There were many different strategies available: you could write a tank that never checked if it was being damaged, but just drove around crazily all the time to be hard to lock onto; you could write a tank that, when it got a ping, would try to lock onto that tank and follow it and keep shooting it until it was dead; you could try to write a balanced tank that would check if it was damaged and evade if so, try to figure out where other tanks were and just send shots in that general direction, etc. We had great fun with it for a while, and then one of the developers (not me, sadly) wrote a tank program that was dramatically more effective than all the others. The fun died away when it became "watch Rich's tank destroy your tank and all the others".
The question is whether Rich's program was actually optimum in some sense (did the best possible according to the simple simulation rules) or whether we could have beaten it if we had been more clever. I'm not sure. I wish I had copies of the source code to all the bots from back then, now that I have a lot more experience in software development and I might get more out of the game.
This was years ago and I couldn't tell you what game it was exactly, but there are plenty of programming games around.
lf(1): it's like ls(1) but sorts filenames by extension, tersely
It may be slightly interesting at first,but the best ones will quickly rise to the top, and no new ones will be capable of competing
I've been watching computer chess since the 80's, and there have been quite a few changes at the top. Any team can have the next brilliant idea.
I'm waiting for an autonomous demolition derby. That might be interesting.
People watch racing because there is risk of a crash with humans in the cockpit.
That is totally absurd. People love watching destruction, yes, but humans do not have to be involved for enjoyment - witness the great ratings shows like Robot Wars got, and those were glorified remote control cars. People just liked watching them violently disassemble each other...
The same will be true of e-racing. Fans will still thrill to a crash, because it will still show basically the same thing - a super expensive car disintegrating into scrap. In fact though it could be even more fun than human races since the rules could be altered such that AI cars had to drive through any wreckage present, no cleanup during the race. That would be awesome to see as AI did high speed moves to avoid scrap...
"There is more worth loving than we have strength to love." - Brian Jay Stanley
If you had bothered to read my whole post you would have seen I mentioned specifically Robot wars. Ratings were mediocre at best on a mediocre market channel. Racing has a massive fan base, sell out crowds across the globe, massive amounts of funding for product spokespersons, massive sponsor contracts and awards.
If human risk was not an issue in drawing and maintaining crowds then you need to explain a continued success of boxing, UFC, Xtreme Games, The Blue Angels and Redbull extreme flying events, various forms of racing. While people don't necessarily root for injury Football, Basketball, Hockey, and Baseball all have large elements of human risk. That risk may not be the whole reason for professional sports popularity, but to deny it's a factor is an outright lie.
-The wise argue that there are few absolutes, the fool argues that there are no probabilities.
Self driving cars aren't going to be terribly good at measuring road feel and that moment when you feel grip suddenly let go and make the correction to stay on the road.
I wish I could see your face when I tell you that the technology to handle those situations has been mandatory in all cars (though not trucks) sold in the US since 2010. It's commonly known as ESP (electronic stability program) and there are a number of ways to actually effect changes in vehicle yaw once it is detected via accelerometer, like decelerating a slipping wheel (with the brakes) or accelerating an opposing wheel (e.g. with an electronic differential and the engine.) Slip can be detected as well (by the use of a second accelerometer) and one common response to slip is to engage traction control, which of course can induce yaw... which is then handled by the ESP.
This stuff began to become ubiquitous in high-end cars around 2000, but it was first brought to the street by Mitsubishi for the Lancer Evo IV and also used on the Galant VR4 and 3000GT VR4, under the name AYC. Even though it was the pioneer, it used the more complex and expensive electronic diff method, which is better than braking because it doesn't slow the vehicle.
All it would take to mess up AI racing is an oil slick or an animal or person or a tree falling or a part falling off another car or any number of other things for the AI to become overwhelmed.
The AI will deal with the oil slick better than a human driving a car without traction control and ESP, because it will effectively implement traction control and ESP. The vehicles already watch for obstacles. It's not that they won't ever make mistakes in these situations, but humans often do as well, so that's not a differentiating factor.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
I'm always impressed by technology when I either see an expert driver fail to achieve something a computer has no problem with, like hill climbing on an icy slope, or your aforementioned oil slick.
But really it impresses me more to think that these days you can just slam your foot on the brake and maintain vehicle control far better than any expert driver on a pre-determined track paying 100% attention ever could. It's also re-assuring going around a corner and seeing the traction control light start flashing on the dashboard before I even realise that there's anything amiss with my car.
Screw humans. We really suck at driving.
But really it impresses me more to think that these days you can just slam your foot on the brake and maintain vehicle control far better than any expert driver on a pre-determined track paying 100% attention ever could.
Indeed. The most modern ABS can even detect when the vehicle is on a loose surface, and lock up the brakes for a moment intentionally to e.g. build up a pile of snow in front of the wheel to offer something to stop against. And traction control can be tuned to permit specified amounts of wheel slip (and differing amounts at different speeds!) so as to preserve sporty feel or to permit a limited amount of digging into the terrain so as to provide traction by removing the loose top material. Ascent control, descent control, automated rock climbing... the things that can be done even with vehicles with internal combustion engines are astounding. When it finally becomes the common practice to connect a motor to each wheel of an automobile, things are really going to be amazing, because they can respond so much more quickly. It takes about 15ms to open or close the spool valve in a modern ABS control unit. But you can change the output to a three phase motor damned near as quickly as you can read the optical encoder that tells you what position it's in. It also brings a previously unseen level of redundancy to the automobile which should pay massive dividends in reliability.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"