Self-Driving Car Faces Off Against Pro On Thunderhill Racetrack
Hugh Pickens writes "Rachel Swaby reports that a self-driving car and a seasoned race-car driver recently faced off at Northern California's three-mile Thunderhill Raceway loop. The autonomous vehicle is a creation from the Center for Automotive Research at Stanford (CARS). 'We tried to model [the self-driving car] after what we've learned from the best race-car drivers,' says Chris Gerdes (who talks more about the development of autonomous cars in this TED talk). So who won? Humans, of course. But only by a few measly seconds. 'What the human drivers do is consistently feel out the limits of the car and push it just a little bit farther,' explained Gerdes. 'When you look at what the car is capable of and what humans achieve, that gap is really actually small.' Because the self-driving car reacts to the track as if it were controlled in real time by a human, a funny thing happens to passengers along for the ride. Initially, when the car accelerates to 115 miles per hour and then brakes just in time to make it around a curve, the person riding shotgun freaks out. But a second lap looks very different. Passengers tend to relax, putting their faith in the automatically spinning wheel. 'We might have a tendency to put too much confidence in it,' cautioned Gerdes. 'Watching people experience it, they'll say, oh, that was flawless.' Gerdes reaction: 'Wait wait! This was developed by a crazy professor and graduate students!'"
Seconds aren't "measly" in motorsports. They can decide an entire season championship.
but where's the video?
brake brake brake brake brake brake
the word is brake
Brakes, not breaks. Maybe it breaks, and that would certainly freak the passenger out, but I sense in this case it brakes. When you're driving at a wall braking lets you do it again, breaking doesn't. Subtle distinction I thought should be pointed out.
(This post brought to you by the collective might of the Oblivious Flaw In The Headline Committee, newly formed to point out the obvious flaw and thereby negating 50% of the discussion dealing with grammar and spelling.)
Please consider this account deleted, I just can't be bothered with the spam anymore.
Turn left
Turn left
Turn left
Kill all humans
Turn left
Not that I will ever trust an autonomous vehicle with my life.
Because you're irrational. I bet you'd trust a human, though!
A self-driving car doesn't have to pay much attention to the fragility of the human form when it doesn't have any on board.
Accelerate at 50g? no problem just add extra bracing.
"In a Race Between a Self-Driving Car and a Pro Race-Car Driver, Who Wins?"
No.
The Internet King? I wonder if he could provide faster nudity.
I'd like to see how this tests out with people other than me and my family as guinea pigs. Will a human driver be able to take back control of the vehicle, say, if the computer malfunctions causing the car to accelerate towards the edge of a cliff? Or will it just say to you, "I'm sorry Dave, I can't let you do that." as it goes over the edge?
There wasn't anything remotely related to the title, no video, no telemetry not even laptimes. And "measly seconds"? Full seconds under racing conditions are not "measly".
I was really disappointed, the title sounded really promising...
Not quite, if I recall correctly, it wasn't behaving as a racer, it just did a GPS-navigated lap of the track -- a human took it for a spin, it recorded the lines and then just replayed them. And it was a regular model (maybe even diesel), not a performance (M-something) one.
Its strength and its speed are still based in a world that is built on rules. Because of that, it will never be as strong, or as fast, as we can be.
Now we just have to see how it handles dogs on the track and the odd drunk or two! A farm wagon pulling onto the road right in front of it would also be a learning experience, to say nothing of the odd whitetail deer!
One more step towards ideocracy and less freedom.
Get used to the idea that in a decade or two's time it will probably be illegal to drive a car manually. New cars might not even have controls, just a microphone to speak the destination into. Mothers Against Drivers, the government and all car manufacturers will successfully collaborate in brainwashing the general population into believing that humans are mentally/physically incapable of actually driving a car at all, certainly never safely.
I bet you'd trust a human, though!
Either way, you're trusting humans. The question is, which choice offers fewer points of failure? ;)
a dozen driverless cars designed to race go at this. Would emergent behavior appear? Can we make them so decisions are recorded and then applied to the next situation?
The Kruger Dunning explains most post on
And the bot still didn't use a turn signal.
what about utility trucks that need to be place that just having a microphone / touch screen will be a very poor way to get them in to place to work on lines / even be used as a temp prop on a power pole to hold it in place.
a race track is a poor test for a day to day use of a car.
It might not be a good test for how many groceries you can carry but its an excellent place to develop and test new technologies and find out what a car will do in emergency situation. If it's raining I'd feel much more comfortable if the car that's driving me has demonstrated an ability to recognize and correct over-steer or know the balance between braking/steering input when a deer jumps into the road.
EHRMAGERDES!
Rainy day one thing.
Rainy day with a bunch of kids, trees jumping out (honest ossifer), ostriches running around, moose attacks vehicle... Would depend on how reliable the system has proven to be.
Software also undergoes thousands of hours of testing before being set loose on the road. Human drivers, not so much.
The problem with utility truck drivers, they won't like not being able to stop at the strip club and bars along the way... They really hate GPS tracking...
It's a BMW, turn signals are optional extras.
Personally, I would LOVE if Top Gear (UK) brought in this team to test how quickly they can make the Reasonably Priced Car go around the track.
For every one of them that drives off a cliff, it'll save 100 people who would have not noticed a stopped vehicle ahead while checking over their shoulder while merging or running a red light because were momentarily distracted, or looked down to check the stereo for a second and drifted into oncoming traffic.
"But only by a few measly seconds."
In auto racing, a few seconds is generally considered an enormous lead.
Obviously I wasn't riding shotgun with a robot. Instead, I was on an airboat in the Florida everglades. The first time the captain pointed us straight at the mangroves and gunned the engine, I freaked. Then I realized that with an airboat, you change the attitude, drift, and then accelerate. This guy was a master. He was able to navigate through the groves with just a couple feet on either side (at slower speeds of course). After the first few high speed turns, I sat back and enjoyed it just like the passengers in those cars.
He took pride in the fact that his boat had no rudder in the water, and thus was less likely to harm the endangered manatees. That first turn around a corner at speed was just one part of the experience. If riding an airboat through the glades isn't on your bucket list, it should be.
For all intensive purposes, "whom" is no longer a word. That begs the question, "who cares"?
I'd be more worried if the computer isn't able to block human input that would lead to an accident. Because that is much more likely.
Humans are idiots. Distracted humans in a reflex situation doubly so.
Well, I might have a way, but it only works on a semi spherical planet in a vacuum.
And, unfortunately, software has little or no responsibility if it gets it wrong while a human does.
Don't think you'll see a CEO jailed any time soon because the company had inadequate testing procedures and they ran a kid over. Money will change hands, regulations will tighten, but people inside a software-producing company going to jail when their product fails? Very rare.
The other thing about humans is they have a superior (if slower) decision making capability. If you've seen the movie I-Robot (hardly Asimov-related, really, but not the point) you are given an example. The problem is that the computer follows orders, which aren't always ideal in all situations.
I'd rather crash into the cardboard box in the alleyway, or mount the kerb (if it's clear) than not get out of the way of the out-of-control juggernaut heading towards me that braking or staying still won't avoid.
That has to be in the top 5 of incorrect word usage; fingernails on a chalkboard. Perhaps TFS needs to be written by bots too.
As self-driving cars age, their reaction times slow down, they leave the turn signal on, and they mistake the farmers' market for the mechanics' service bay.
The human brain goes through a little over 140,000 hours of development and testing before being set loose on the road.
-
If a human driver accidentally runs over a kid who jumped out on the road, they don't get charged with a crime.
In the iRobot example, the robot saved the person they perceived to be the most likely to survive. A human would have tried to save the child and the more likely outcome would be two deaths instead of one.
Its a pretty slow learner. It takes over 10,000 hours to learn how to walk. After 20,000 its still learning fine control of individual fingers and still doesn't speak fluently. 0 of those 140,000 hours has been spent driving a car. 0 hours in a simulation, 0 hours learning how to correct an out of control car, 0 hours in emergency situations.
Don't think you'll see a CEO jailed any time soon because the company had inadequate testing procedures and they ran a kid over.
I don't think so either. But then again, I also don't think he should.
How worried are you that a bridge you drive over won't fall if the winds pick up. How worried are you that the skyscraper you're on in LA won't collapse to the ground if an earthquake happens? These are all real dangers. And an engineer signed off on the design, indicating it meets all of the applicable codes. If something happens as a result of negligence from that engineer, his ass is on the line.
I am an engineer, and I know for a fact I'd face jail time if something happened to anything built from one of my designs that can be traced back to my negligence. I don't see why it would be any different with these cars.
The other thing about humans is they have a superior (if slower) decision making capability.
Not always, no. I've seen humans make plenty of bad decisions while driving that I know for a fact would not be made my software.
I'd rather crash into the cardboard box in the alleyway, or mount the kerb (if it's clear) than not get out of the way of the out-of-control juggernaut heading towards me that braking or staying still won't avoid.
And why are you assuming software won't? Those are not difficult human-only decisions, they're pretty standard and easy to recognize things. Currently google cars will actually avoid pedestrians that jump in front of the car at the very last minute. I don't know many people who I'm confident would be able to avoid hitting the pedestrian.
Furthermore, a driver goes to jail because he killed someone while driving drunk? Tomorrow another driver does the same thing. A car kills someone because of a software error? Next week, every car has received a software update that assures that error will never happen again.
According to this source USB can get down to around 1 ms theoretically. In practical terms, you can get an expensive MIDI keyboard with input through USB, and that's certainly much faster than 90ms, otherwise it would be completely useless.