Slashdot Mirror
Racing a Real Car While Wearing an Oculus VR Headset (wsj.com)
pbahra writes: In a race that fuses video-game technology and real world driving skill, two professional drivers, on two separate but identical tracks, have raced against each other — effectively blind — while wearing virtual reality headsets attached to their crash helmets. The drivers hurtled around the circuit in two identical 2015 V8 Ford Mustangs, trusting that what they were seeing on their Oculus Rift DK2 VR headsets was a true, real-time representation of how their cars were performing on the actual track. One of the main challenges: tracking the cars' exact positions as they sped around the track without the need for re-calibration. This was necessary so that an exact match could be achieved between what was happening on the physical track and its representation on the VR screens.
Real Men wouldn't have been on separate tracks.
This assumes that what you're seeing with your eyes is "reality". There's no absolute proof of that. :)
"I have never let my schooling interfere with my education." - Mark Twain
How do you do that? I should think building a second track which is identical to the first would be a hell of a feat.
All it would take it relatively small differences in the track and it's going to make a huge difference.
I've never heard of such a thing, any civil engineers who could tell us how hard it is to have two identical race tracks? I just can't see it being easy to get the same grading and all that in two separate places.
Lost at C:>. Found at C.
See, to me, this is bass-ackwards. The point of the autonomous automobile is to remove the driver from the equation of basic transportation, or to at least have the option to do so when one doesn't feel like driving.
This thing doesn't make a whole lot of sense to me. Generally the point in sport driving is to enjoy the drive itself, and to really connect with the machine where it feels like an extension of yourself. Removing that experience in order to simulate it in a 3d shell while still occupying the vehicle makes no sense to me. If you're in the vehicle you'll probably use your own eyes to operate it, or you'll use less expensive vision augmentation technology where your eyes are not adequate, and if you're not occupying the vehicle then you'll probably still get enough bandwidth out of video transmissions from cameras to allow you to operate without needing to encode the experience and then generate that 3d quasi-virtual environment.
I don't really understand what this particular technology is for.
Do not look into laser with remaining eye.
I would call real-time, high speed driving relying entirely on the VR is fairly impressive in terms of the quality of the system.
If they can navigate a real car around the track, networking people into a simulation is probably much easier.
It's fairly cool, and involved a lot of technology. It may end up just being a PR stunt, but such technology has a way of having someone say "hey, wait a minute, if I had one of those, I could ..."
I should think being able to do this and have it work means you are tracking the real car and the VR car exceedingly well ... which suddenly means there's probably lots of places where remote operation becomes possible.
Lost at C:>. Found at C.
c'mon, man!
Use The Force, Luke.
Let go.
Luke, trust me.
https://www.youtube.com/watch?...
They also had the same problem of tracking the car exactly as they found that even the difference between front and back seats would cause nausea.
bickerdyke
Then why are the drivers inside of the vehicle?
The only kinds of places that I can think of that we can't put cameras into and expect them to reliably work are in underwater applications and in extremely harsh post-accident environments like nuclear power plants that have suffered catastrophic failures. Anything beyond these terrestrial applications isn't going to work with this technology because of the latency (ie, can't virtually drive a Martian rover) and while there are submarines and robots for exploring Chernobyl sites like Fukushima Daiichi, the applications for those environments are usually based on slow, methodical processes, not on fast-twitch reactions.
Admittedly if the technology to make the interface work in real-time is actually working then that's pretty neat, but at the same time I still wonder what it's actually for.
Do not look into laser with remaining eye.
See, to me, this is bass-ackwards. The point of the autonomous automobile is to remove the driver from the equation of basic transportation, or to at least have the option to do so when one doesn't feel like driving.
Unrelated to TFA and your point about sport driving (which I agree with completely), the current goal of autonomous vehicles is backwards imo. Rather than shoot for sensationalism (full autonomy), they should be going for things augmenting driver abilities. Lane assist, brake assist, adaptive cruise control, traction control, active stability system, etc. As you integrate those technologies and give them more and more control over the car, what you'd end up with then is a car that basically is driving itself but with the "driver" making the decisions. Twitch the wheel and the car automatically signals, checks its blind spot, then moves over a lane if safe, at which point the driver goes back to reading the morning news as the adaptive cruise maintains safe following distance (pair the system up with a tablet and it could overlay lane status so the "driver" maintains some degree of situational awareness as fed to them by the car). The car could even alter driving modes based on current circumstances, e.g. upon exiting the freeway, the driver is given full manual control until they enter a parking lot and attempt to pull into a space, at which point the car detects it and offers to take over.
They could have, but that's what they normally do, so what's the fun in that.
This was an engineering challenge, in terms of latency in particular, and I bet it was fun for the drivers too to have the CGI racetrack.
Then why are the drivers inside of the vehicle?
Because the sponsors are castrol? [see the placement at about 1:15 -- seems they are pouring castrol into the simulator car which is anyway standstill]. May be if it were a tech company, the drivers will be in a data-center kind of environment.
Actually this technology if it could be implemented reliably could be very useful. Imagine if you could have a full heads up display where there's no blind spots coupled with self-driving technology to alert you visibly on the screen if you were about to slam into another object or point out potholes, ect with driving guidance. I imagine the military might want something like this as well as the windows usually arn't as well armoured as the vehicle itself or there's a tonne of blind spots in an actual tank.
See, to me, this is bass-ackwards. The point of the autonomous automobile is to remove the driver from the equation of basic transportation, or to at least have the option to do so when one doesn't feel like driving.
...they should be going for things augmenting driver abilities. Lane assist, brake assist, adaptive cruise control, traction control, active stability system, etc...
Which one of those aren't implemented yet?
As for the rest of your point, I disagree inherently. I don't trust people to be smart enough to know when they should "Twitch the wheel" and then go back to reading the news. I know too many people who can't even tell they haven't hung their phone up for me to be able to trust that they've let the system properly take control.
"Don't meddle in the affairs of a patent dragon, for thou art tasty and good with ketchup." ~ohcrapitssteve
Yeah, the states that I want an autonomous car to operate in would either be fully autonomous where the car figures out the particulars of where it needs to go, or occupant-instructed, where the car can be given directives to try to follow ("keep left" or "slow down and keep right" or "pull over in the next parking area"), but neither require the car to be operated by the person. This would let a child or someone that does not have a license use the car to get from place to place including establishing reasonable curbside service or the ability to make endpoint-specific decisions.
I would love to have a car capable of sport-driving that could also be fully autonomous. Sometimes I want to enjoy the drive, other times I want to enjoy the view.
Do not look into laser with remaining eye.
Possibly not the same result. The VR headset only supplies visual information, and possibly auditory information as well. In driving -- especially racecar driving -- the tactile and proprioperceptive (i.e. acceleration) data inputs play a big role. If they had a remote driver, it's more likely there would have been problems. Or at least, the drivers wouldn't have driven exactly the same way, since they would be driving "blind" in a more profound sense than the summary implies.
Any sufficiently simple magic can be passed off as mere advanced technology.
90% of my driving is not especially enjoyable. Being stuck in traffic or driving for hours at constant speed on straight lines while still needing to maintain attention is not my idea of fun.
And the remaining 10% are not pure bliss either. The public road is not a toy, and if you "enjoy" it too much, cops will remind you about the rules, or worse.
And if I am on the track, there is no way I want any VR stuff. If I pay good money for the real deal, this is not to get some glorified video game.
Though I agree, it was a PR stunt. :)
Any sufficiently simple magic can be passed off as mere advanced technology.
Everything is being Recorded --- that's the point. There will also be driver-less ("A.I." only) Nascar races. The data from these kinds of experiments is going to be extremely valuable. I would lay odds on Nascar being completely driver-less --- as far as having a physical driver in the car within the foreseeable future.
While I expect that you are correct about the non-visual feedback some of this can be simulated e.g. by tipping the drivers chair back for acceleration. While this might not be perfect I would love to see what a Formula 1-style race would look like if you had the drivers sitting safely in chairs so you could remove all the rules which slow the cars down to make it safer for a driver in the vehicle.
Sounds interesting.
Would be great if they said:
- how it worked
- did it work?
- what speeds did they achieve
- driver's opinions
- surprises?
- how did they cope with in-race obstacles? Could they just drive over them?
- same with the other car, a HUGE part of racing is positioning...in this could they just drive through the other car?
BUT THEY DON'T
Worthless fucking promo video.
-Styopa
Which one of those aren't implemented yet?
They are implemented, that's the point. What I'm saying to do is implement them more. e.g. lane assist generally just beeps at you or vibrates the wheel on the side you're drifting. The next step would be for it to actually control the steering unless you explicitly override it. Blind spot detection also just beeps at you if you signal and/or drift lanes while there's something there. The enhanced version just flat out wouldn't let you do it.
As for the rest of your point, I disagree inherently. I don't trust people to be smart enough to know when they should "Twitch the wheel" and then go back to reading the news. I know too many people who can't even tell they haven't hung their phone up for me to be able to trust that they've let the system properly take control.
Yet you trust that they'd let a self driving car take control? (assuming that self driving cars have a manual function, which would be asinnie not to include). The whole point of the system I was describing is that the car is in control, it's just that the car has no executive function and allows the driver to guide it. If the driver were to bail out while swerving down the freeway at 80mph, the car would revert to a basic lane following behavior, presumably coming to a stop once it runs out of freeway and encounters another car at a traffic light (at which point the adaptive cruise control would perform an emergency braking maneuver so as to not hit the car in front).
Think of it more like a self driving car in which the user can press buttons to give it turn by turn directions.
I would love to have a car capable of sport-driving that could also be fully autonomous. Sometimes I want to enjoy the drive, other times I want to enjoy the view.
I feel like those two things are mutually exclusive (depending on your definition of sport-driving). All of the things required for a sporty car (small, lightweight, lots of internal bracing, good balance, as much power from as little engine as possible) conflict with all of the space and weight considerations of the sensors/electronics/actuators that would be required to make something fully autonomous.
I donno, I have an awful lot of fun in my '95 Impala, and it's bigger than and just as heavy as anything else on the road today short of a full-sized truck.
Do not look into laser with remaining eye.
You'd be better off removing the polarized film from an LCD panel and installing that film on a pair of glasses. You'd have a monitor that only you could see.
Do not look into laser with remaining eye.
They are implemented, that's the point. What I'm saying to do is implement them more.
Let me guess: You're some kind of well-paid manager somewhere.
Take your ABS-equipped vehicle out in the snow. Go fast. Stomp on the brakes. The car just refused to perform an operation - it did not lock the brakes when you told it to. It did not apply the full force of the brakes as instructed.
For the record, being able to disable ABS for certain driving styles would be enjoyable for me but I'm glad that people have access to them.
"So long and thanks for all the fish."
That is probably what consumers what, but the goal of autonomous cars is much grander than that. They market it as driving for you as a way to sell it because safety sells, at least when it comes to cars. The goal is actually to network transportation to make it much more effective, so instead of having 1 self-driving car, the whole road is self-driving. This way they can route around congestion, bad weather, construction, etc.
I am much more interested in this kind of thing than in Google's nonsense with self-driving cars. I understand Google is paving the way for improved versions of Tesla's autopilot so it is great to see, but as far as personal enjoyment of driving an augmented-reality experience will be the future and bring great benefits.
Yeah, it's great to see the driving experience being augmented for a change, instead of being dumbed down, which seems to be the norm these days. Still, I can see a lot of value in self-driving cars, even for private use, under certain circumstances or for some people.
I still wonder what it's actually for...
Entertainment is always an option, even if nothing else comes out of it. Just imagine playing something like Battlefield YOURSELF, like in paintball style, but wearing VR glasses instead of a protective mask. All of the needed ingredients seem to be maturing, like VR displays and cameras, motion control, etc. I can almost see myself moving to Battlefield 1942 Desert Combat's Al Nas or Lost Village... I could finally get off the damn chair!
Then why are the drivers inside of the vehicle?
1. No FBW steering or braking.
2. Non-visual senses such as acceleration and sound are very important to performance, especially on a racetrack.
It is dangerous to be right when the government is wrong.
I still wonder what it's actually for...
Entertainment is always an option, even if nothing else comes out of it. Just imagine playing something like Battlefield YOURSELF, like in paintball style, but wearing VR glasses instead of a protective mask. All of the needed ingredients seem to be maturing, like VR displays and cameras, motion control, etc. I can almost see myself moving to Battlefield 1942 Desert Combat's Al Nas or Lost Village... I could finally get off the damn chair!
This.
The headset merely replaces the real boring visuals with CG, so instead of seeing your green paintball explode on your mate's pokemon tshirt instead you see a bullet explode through their high tech chest armour, emblazoned with their team's emblem, and if they get up again afterwards and try to shoot you, it won't matter so much, aside for the paintball impact, they'll still appear dead to you and you can continue playing. They've already got a few places experimenting with this sort of thing (One in Melbourne, Australia, that I might go check out this christmas)
"lt;dr" is the correct response to most of my posts.
You *cannot* refuse to perform a commanded operation when the driver is in control of the vehicle.
Yes you can. ABS refuses to stop if it senses the wheels locking up. Traction control refuses to give full power to a wheel if it senses slippage. There's no reason that the blind spot detector can't refuse to let you switch lanes if it detects a motorcycle there.
And if flat out refusing the command has you worried (in which case, why don't fully autonomous cars also have you worried?), it could just provide the refusal as making the steering wheel hard to turn in that direction rather than flat out locking it.
Then why are the drivers inside of the vehicle?
The only kinds of places that I can think of...underwater applications...nuclear...failures...the applications for those environments are usually based on slow, methodical processes, not on fast-twitch reactions....Admittedly if the technology to make the interface work in real-time is actually working then that's pretty neat, but at the same time I still wonder what it's actually for.
The military has already been researching and investing in similar technology for vehicles with limited views and/or armoring, like tanks, trucks, helicopters, planes, etc. Transferring this tech to entertainment applications was already stated.