I'm sorry, but omnidirectional treadmills can't work and here's why | Hacker News Reader

I'm sorry, but omnidirectional treadmills can't work and here's why

james.darpinian.com

72 pointsmodeless58 comments

As stated in TFA, what if the holodeck space were larger, e.g. 5-20 meters? The centering adjustments can then take place gradually (unless you flat out sprint straight like Usaine Bolt). It kind of irks me this is a valid solution which is half glossed over to make room for a clickbait headline.

This could ostensibly function as a thorough bypass to compensate for the sensitivity of the human vestibular system, at the cost of TAM (Total Addressable Market, due to size constraints).

Personally I'd be up for going to a Lazr-Quest like place to experience a real-ish holodeck. Would make for a great destination for a venue in SF.

Related submission from yesterday:

Disney unveils the HoloTile floor

https://news.ycombinator.com/item?id=39156444 (288 comments)

And the direct link to the goods: https://youtu.be/68YMEmaF0rs?t=203

Then you have to find someone to make/modify your favorite action game that's compatible with this, which is no easy task in and of itself. Ladders and clambering you experience in game will still be a matter of pushing a button on the controller in your hand.

This is a problem for once the thing is working ;)

No doubt there'll be other constraints coming. This isn't The Ultimate Solution, rather a significant incremental improvement over the stated hurdles with existing designs optimized and marketed towards fitting into 3'x3' space in front of your television, bed, or desk.

This will be pretty trivial. There is very good support for common apis.

The article addresses this exact question.

Though the comments also address using boots instead of an external device, but then don't consider that in the context of the arguments against a larger space.

Mostly it says it’s hard and the machine would have to be big. So?

Same for galvanic whatever. “It would be hard to make it so precise as to not make you sick”. Ok, try hard then.

Okay, let's think this through.

- a large area is necessary

- the machinery is going to be extremely expensive

- it's still going to be a subpar experience as the virtual area will never be fully represented unless you remove vegetation, cliffs etc

- it needs to be supported by the software (which nobody will, because there is no market for it)

If you now make a venn diagram that represents all people potentially interested in this, you end up at pretty much 0.

Thus, it doesn't work.

More points.

- our equilibrium sense in the inner ear is very sensitive to velocity changes

- the machinery only fools for steady velocities or small velocity changes

- turns in walking imply strong velocity changes

The effect is that we tend to counterbalance the velocity changes but because the machinery cannot simulate velocity changes, counterbalancing makes us fall. If that gadget manages to sell expect many hilarious videos of people falling on their noses or butts. Exactly like for the treadmill. Note how people always fall if they assume a velocity change (like starting to run by jumping on a running treadmill). I always laugh when the treadmill then sends people off. It's hilarious.

Perhaps with some tilting platform the velocity changes can be emulated. Whenever the person changes the direction of the walk, the tilting platform tilts in the opposite direction of the velocity change. Let's say a person walks to the right and suddenly walks to the left. The velocity change vector points to the left. The platform will tilt to the right. This is difficult to get right. Probably latencies of more than five milliseconds will irritate our equilibrium organ I think. Software should be fast enough but the platform has to move swiftly.

And even this is not enough. There will be rotation in the tilting. Some sensitive people will feel sick.

And the quickly moving platform is dangerous. A bug (wrong sign of the vector) and the platform kicks people off it. Or some body part gets below the platform and the crush is strong enough to grievously hurt people.

> And the quickly moving platform is dangerous. A bug (wrong sign of the vector) and the platform kicks people off it. Or some body part gets below the platform and the crush is strong enough to grievously hurt people.

Ski lifts are a thing. People mostly are fine with the risk.

All these comments are basically nomad level rejections.

This reminds of the dude who wrote a multi-thousand-word analysis bashing the Apple Vision Pro and how such device shouldn't be able to work because Apple didn't get it right... without even holding said device in his own hands, let alone test it.

Why??

Didn't smart people predict, time and time again, that some new tech would never work because it goes against all rules of common sense and "physics"; only for the people building it to defy all these "rules" and turn their crazy idea not only into a real object, but also into a product that can mass produced and used by millions?

Can't we just learn??

I think there might be a problem with calling these new VR systems are 'holodecks', a term used both by journalists and sometimes the technologies' inventors themselves. Just because we haven't yet reached a USS Enterprise level of refinement, in which you can run, climb and interact with objects naturally, doesn't mean that omnidirectional treadmills don't have a huge amount of potential. After all, I for one don't feel any less impressed by ground-penetrating radar or X-ray scanners just because they won't fit in a pocket-sized Star Trek tricorder. All that will come in time, probably within my lifetime. That's already a century or two ahead of their fictional schedule :)

We could use omnidirectional treadmills right here and now, the vestibular confusion issues notwithstanding, for all kind of entertainment and education purposes. Consider that a lot of popular video games don't actually involve as much complex movement as you might assume from the critiques of VR: the main activity in most tactical first-person-shooter games is crouching behind walls! Virtual reality museums and art galleries will be great with omnidirectional treadmills; here, walking is not the point, it's just a natural and intuitive way of navigating in a virtual environment without having to learn the controls beforehand.

I get the point that omnidirection treadmills "can't work" in the sense that they can't be a perfect simulation of real walking, but I'm not sure they need to be perfect.

Current generation VR systems are highly flawed and far some perfect, but they offer something very compelling and enjoyable.

I don't know how good an omnidirectional treadmill needs to be before it crosses the "good enough" threshold, but I hope people continue to build and research these to find that threshold

The problem is that they basically all provide negative value.

They cost a lot, they take up a lot of space, they get in the way while using VR, and they don’t do anything to solve motion sickness. The situation is not that they aren’t perfect yet, but that you would be better off not having it at all.

I think this take is correct and lacking a “jacking in” to your neural cortex a la Neuromancer or The Matrix, true VR will remain a pipe dream.

I find it very sad that someone thought about this long enough to write an article without considering "What if you just get used to it?" Like fuck me, people train to put their bodies in an enormous variety of physical circumstances. If a human being can learn to drive a car at a hundred miles per hour, can learn to fly a wingsuit, can learn to exist in zero-G, they can probably damn learn to walk on an omnidirectional treadmill without falling over or throwing up.

You make a good point -- basically, our brain could understand that our body is fixed in space, and that when we "walk" we're actually sliding the environment around with our feet. And just quickly get used to it.

It's not clear that this would work, but it's not implausible either. Maybe it'll make us nauseous, or maybe it'll feel different but still work fine.

The penultimate paragraph addresses this point.

> [...] Maybe we can still use them unnaturally. If you train yourself, maybe you can learn to do the strange movements that are required to change speed and direction on the treadmill. [...]

But that doesn't address the neuroplasticity that's possible with this kind of learning. It's not just a matter of training yourself, is that, when you train yourself, the movements become natural, and feel like a reasonable way to move yourself through space.

This is the most doomer take on the problem. You could say the same thing about learning to write or do math.

Sure, but it's not a great simulation of reality (the R in VR) then now is it?

You'd be surprised at what the human brain can accommodate for. Most long-time VR users will be able to tell you about instances of the VR causing phantom sensations of touch that come purely from visual stimuli. The human brain does a lot of post-processing of sensory input to conform to its predictions - there's a huge gap between what we think we feel and the actual inputs. Even fancy haptic gloves which only add crude sensation can provide almost lifelike sensations when combined with visuals.

Since when has "being indistinguishable from reality" been a requirement of VR?

In that case you can just get used to a joystick or some other input method.

The whole point of the treadmills was to avoid having to get used to it.

That's like saying if you have sex with a condom you might as well fuck a wooden doll.

I don't get the obsession to reduce such an obvious continuous solution space to literal black and white answer. Imagine if doctors did that. Oh this medicine only gives an additional 20 years of healthy life but doesn't cure the illness, oh well might as well not do anything then.

No, because the treadmills do literally nothing but get in the way. They don’t reduce motion sickness, and they don’t enhance the experience. And it seems unlikely that it’s even possible to solve that problem.

You would have a better experience using a joystick and just getting used to it.

They do not reduce motion sickness. But they for sure enhance the experience. It's simply incomparable to walk yourself vs using your thump to move. I can't fathom someone trying an omnidirectional treadmill and coming away with opinion they are on the same level of experience as using a stick.

It might not ever work "like walking", but nor does a linear array of a bunch of miniature speakers in a pair of headphones work like spatial audio in the real world, but the brain figures this out fairly quickly. The question here is, and in some way the "art" or the creative side of the engineering puzzle, is when does this "feel good enough" for most users?

Probably wouldn’t be the first that can’t happen that’s happened.

We even have a saying for it:

There's a first time for everything.

Tilt it? That is to say, to some extent, could you repurpose gravity to give a sense of acceleration?

> blah blah .. Vestibular system .. blah blah

Sorry, but the reason they don't work is much easier to describe using momentum. You can't stop on a dime, turn left quickly, or any "regular" human movement that alters your momentum, because you (~50-80kg) aren't moving*.

Even if we HAD no Vestibular System, this would still not work as expected; we fall over the moment we try to "stop" while running at full pace.

If you are interested in this you may also like the "Exit Suit" which is open sourcing their plans at the end of this month.

https://www.youtube.com/watch?v=fZiwrDfi9rw

Is there a way to control the pitch and roll of the treadmill to simulate an acceleration change? If the vestibular system is expecting a deceleration, make the treadmill have a slight upward slope along the axis where we would expect to slow down?

Even that might be a bit overcomplicated, you could probably get quite far with a bowl shaped treadmill (I mean, anecdotally the simple wheel shaped ones do feel like running, their inertia notwithstanding).

If you start pitching the treadmill the tricky part is that you'd probably need to rotate the person around their centre of mass, or somewhere close to it.

I like how even in the advertisement in the video it is perfectly obvious that the act of walking on this thing must be quite unnatural. The person using it looks like he is half way though the first day of walking school.

riding a horse doesnt feel exactly like walking so it's a doomed technology

Those that say a thing is impossible should stop interrupting those who are doing it.

Meh. Author's not wrong but their their criticism may or may not apply to you individually. Different people experience different levels of vestibular decoherence.

It may impact the marketability of such a device if only 50% of the population can use it without getting really sick, so that matters a lot. But whether an individual is susceptible varies from person to person.

I find it extremely disappointing, when a group is attempting to innovate on a problem that is very clearly difficult, and someone comes along to say "I'm sorry, but this can't work."

The researchers are obviously aware of the challenges and working to overcome them. No one is confused to think that an omnidirectional-treadmill can capture the true physics of walking. The researchers are working to get it as close as possible.

Naysaying, in cases like this, is not a virtuous skill.

I don't even think this is the biggest problem in VR, and it's mostly solved.

I used a Quest 2 for the first time the other day. It was a very simple environment, a single room you could walk around in. There was a desk in the middle. I tried to lean my elbow on the desk and almost fell over :/

So kudos to the Disney team.

Strong agree. It’s boring, lazy, and wrong.

The Apollo program is notorious for inventing all kinds of technologies as part of the effort. We should celebrate ambitious “moon shot” (or loon shot) efforts.

It’s awesome when companies support ambitious and crazy efforts. These projects often pay all kinds of unexpected dividends. In both technological development and people development.

Was landing on the moon useful? Not really, no. Was the effort to land on the moon useful? Hell yeah it was. Kudos to all the companies that fund bonafide research.

One could construct a spectrum, from outright pessimism through 'naysaying' and 'reasonable skepticism', ending up somewhere in the region of constructive challenge and helpful pressure-testing discussion.

Would all stations on this spectrum be disappointing to you, or is there a line to be drawn? And if so, how would you explain your reasoning?

---

(I ask this not to be awkward, but because I've often mused over something similar myself, when frustrated by the knee-jerk challenge offered by some folks to new or positive ideas.)

I agree, it's certainly a spectrum.

Theranos comes to mind as one end of that spectrum. But in that case, the naysaying was around the physical impossibility of their core proposition. And, that was not a case where "good enough" was fine. The fact that they kept their supposed invention secret was another good reason to naysay.

Here, Disney is demonstrating their result and progress. I didn't see if there was a research paper attached, but I would fully expect it to openly discuss the limitations.

I think they are talking not to Disney Research, but to the people excited by Disney Research. I’m of the opinion that we should just let people be excited about things, but I can see the mindset that doesn’t want people to be set up for disappointment.

People Who Say It Cannot Be Done Should Not Interrupt Those Who Are Doing It

https://quoteinvestigator.com/2015/01/26/doing/

This is the "stick my head in the sand" techno-optimist take. The physics of our world defines what is possible, and no amount of "trying to innovate" can overcome, for instance, the poor economics of the Hyperloop or flying cars. This author makes a compelling argument that the physics underlying the human vestibular system is unable to handle the Omni treadmill, even with a technically perfect implementation. You can't just "innovate" past physics.

Don't use poor economics with physical impossibilities interchangeably.

It's something that my late ICE professor did all the time about EVs in the mid 2000s up until his death in the mid 2010s. He was extremely bright, very knowledgeable and pretty much "the" in his little ICE niche and yet he was always talking about how EVs wouldn't be teh future because [physically possible but outlandish thing]. Energy density, price per kwh, weight, battery management, charging infrastructure. All things that he was all to happy with going on national TV to declare that EVs were a pipe dream. Unfortunately he died at a young age and didn't get to see the EV revolution and the Tesla Model Y selling more units than his beloved Golf in Europe.

The thing is, and that's my actual point, you can't innovate past physics, but you sure can innovate past poor economics.

It could have hedged a bit, but the article is a pretty detailed explanation of the challenge from someone who has used one before. It's useful information to people who haven't thought it through.

So yes, I'm sure the Disney researchers are aware of the problems, which makes me wonder what they're really up to. Is it just another rigged demo? Will they ever ship?

That's what Theranos said

I have to agree. For me it’s the headline that particularly bothers me. It’s clickbait and cheapens the real points they are making.

The real problem here is how the media has latched onto it. When Disney showed it, it felt like they showed and talked about it in an intentional way. People took it and severely exaggerated what it can do.

Until we see this technology implemented in something we don’t know what they are really intending to simulate here.

I find it extremely disappointing when someone who clearly knows what they are talking about tries to do a bit of a reality check on something that is clearly being uncritically hyped by the press - and then instead of someone actually refuting their arguments they get accused of "naysaying".

Yay.

As if laws of physics or biology cared about our attempts to circumvent them or trying really really hard/throwing money at the problem. All that it takes is to work hard and sufficient media hype, right?

That's not how things work. E.g. people are trying to build "free energy machines" and perpetuum mobiles since centuries ago - so schools teaching that it is not possible are also "naysayers"?

I happen to agree with the author of that blog. I have worked in that field for over 20 years, have tried several of those treadmills (probably most of the ones that made rounds in recent years) - and no, they really don't work outside of very niche applications.

Sure, one can make them physically work (aka it somewhat lets you move around) - but the movement is very clearly unnatural to the user, you have to "learn" how to walk on them (and get on and off safely!). Without a restraint of some sort (harness, frame, railing, etc.) they are at a major risk of injury too. Running or jumping on these devices is usually a quick way to the hospital.

It is simply not at all a practical solution to locomotion in virtual reality.

The funny thing is that the Disney project has zero to do with VR and has not even been mentioned in such context in the original article. It is from the "Disney Imagineers" - the folks developing their amusement park rides. There such device has a clear application potential and is certainly a cool piece of engineering.

Reply on news.ycombinator.com