Daydream Labs: Teaching Skills in VR | Hacker News Reader

Daydream Labs: Teaching Skills in VR

blog.google

92 pointsjanober23 comments

"no matter what warning we flashed if someone virtually touched a hot steam nozzle, they frequently got too close to it in the real world, and we needed a chaperone at the ready to grab their hand away. "

Obviously they need to "die" in game, a few times starting the whole thing all over from the beginning and they'll never "burn" their hands again. I wish I could tell the author this....

> And no matter what warning we flashed if someone virtually touched a hot steam nozzle, they frequently got too close to it in the real world

Yep. I don't think they were trying hard enough to make the experience unpleasant and memorable. If they didn't get further than flashing warnings with various sizes and wordings then they were a little premature in this conclusion.

And the comment about hand tracking is odd in that they haven't even acknowledged the existence of Leap Motion - which would work fairly well in this scenario (less of the occlusion issues that prevent it being a good general purpose VR controller.

>Leap Motion

it's really crappy for VR. You need LOS to the base station, and obscured digits don't show up. Also, the range sucks.

They make a mount for sensor that puts it on the front of the headset and the software corrects for headset position.

The newer model Leap Motion has 180 degree field of view so essentially - if your hands are in front of you then it can track it.

The range only needs to be as long as your arms.

Some hand positions are self-occluding but this is a problem less often than you'd expect.

Direct sunlight used to be a problem because of IR interference but this might have improved (I haven't tried the it since last October).

I own a Vive headset; I do not have a leap motion.

The community opinion on leap seems to be negative. I was basing my argument on that.

Leap motion is so much more natural than any hand held controller its crazy! Using it gets rid of all of my motion sickness.

Is there some reason you are excluding mounting the Leap camera on the front of your headset?

A virtual person staring at them probably would have done the trick. People engage with avatars for some strange reason and I probably could have predicted their outcomes to each of these approaches.

It's already known how to make someone really, really dizzy using VR. I bet we could use that to mimic "pain" and have the brain make sure you don't touch that hot nozzle again.

I think the aim is to avoid people hurting themselves rather than simulating it.

This will be extremely useful for teaching how to use 'always on' equipment like big electron microscopes where the cost of training someone is enormous if you have to give up time where the equipment could be running.

One of my 'someday' projects is a virtual electron microscope ... so much with TEM becomes easier with hands-on experimentation, but that often isn't practical (or even safe for the instrument without someone present to guide you).

Trouble is, something like that requires a fair amount of domain knowledge to code, and the intersection of people who can code and who also understand electron microscopy sufficiently well isn't that large... but considering the cost of the instruments, it'd surely be worth doing.

You should check out labster. They have been working on this for a long time.

https://www.labster.com/

This is funny. All I would've thought you'd need is a score, which you'd increment if you get a shot of espresso out quickly, or decrement if you touch a hot pipe. Let people figure out the rest!

Make it boring. Touch the hot pipe and you have to call an ambulance, wait for the ambulance, go to hospital, wait hours for a doctor (there are more urgent cases...) get bandaged, have the cost of your medical care deducted from your wages etc.

"You may touch the hot pipe as long as you get much faster."

Come on. Modelling hard constraint with a score is tricky.

Anyone have a link to an academic-like paper? Would be interested to see the details of the experiment (i.e % improvement with VR). We think so much about teaching computers it's interesting that we still could have room for improvement figuring out how to better teach other humans. Could we get 2x, 5x or even 10x better at learning? What would that look like?

Something like this for changing drum brakes in a car would be great. Or changing a baby's diaper.

It's probably not worth a headset for any single one of those tasks, but for one that can be used with some training platform to learn any manual task, that would be great.

If they married teaching skills in VR for training and their Google Glass enterprise that would be interesting. Train people to get the basics in a VR that could potentially be done anywhere. Then augment them in the Enterprise with their AR offering Google Glass enterprise.

> In the end, it was much easier to model the trainer like a video game, where every object has its own state.

Uh? Making an interactive simulation was easier if made as an interactive simulation? How could this possibly be a surprise?

It's not like we don't have simulators - flight training being the canonical example.

Is there somewhere I can download this if I own a Daydream headset and Pixel?

I happen to be in the market for a home espresso machine. I'm also interested in learning the "craft" of it, and this could be a fun intro.

This would make an awesome game!

Check out "Job Simulator." It's meant to be entertaining and not teach any real skills, but bears some similarities to what the article describes. My daughter loves it.

Reply on news.ycombinator.com