Forum Discussion
Why ODT’s Do Not Support Natural Locomotion
A linear treadmill provides an excellent simulation of running or walking at a constant velocity, i.e., when no acceleration is taking place. The problem comes when a person tries to accelerate (/decelerate).
To accelerate along a straight line in the real world, one must apply force to overcome the body’s physical inertia (Newton’s laws of motion). All of your reflexes are tuned to apply just the right amount of force to attain the desired change in velocity. Your inner ear senses accelerations, and your proprioceptive system senses the applied and reactive forces your body experiences.
On a treadmill, however, you’re not really going anywhere (or at least not very far). That means you do not have to apply horizontal forces to overcome your body’s inertia to accelerate (with respect to the treadmill’s surface). An ODT simulates what it would be like to remain under the influence of gravity, and retain angular inertia, but eliminate linear inertia. This unusual condition is at odds with a person’s natural reflexes for directing their body’s motion, rendering their natural responses ineffective, i.e., making it difficult to stop, accelerate, or redirect one’s motion quickly on a treadmill.
ODTs are challenging to build because of the demands of their mechanical and control systems. I suspect that a lot of people believe that it is just a matter of perfecting ODTs to keep the user nearly centered of the treadmill in order to provide a realistic simulator for natural human locomotion. That is not the case.
There is a way to simulate linear inertia: Strap the user to an active mechanical tether that pulls and pushes on them to apply the forces necessary to substitute for the (simulated) absence of inertia. The Sarcos Treadport system did just that, albeit it is a linear treadmill:
http://www.cs.utah.edu/research/areas/ve/LocomotionDisplay.html
I have yet to see such a harness built to apply omnidirectional (horizontal) forces for simulating inertia on an ODT.
There may be another alternative. People can learn a wide variety of motion skills. They can learn to skate and ride a surfboard. With enough practice people may be able to master virtually accelerating and decelerating on a well-designed ODT. But they will not be performing natural locomotion and there is no reason to believe that their performance (speed & accuracy) at varied motion tasks will match their real world performance.
-JimT
To accelerate along a straight line in the real world, one must apply force to overcome the body’s physical inertia (Newton’s laws of motion). All of your reflexes are tuned to apply just the right amount of force to attain the desired change in velocity. Your inner ear senses accelerations, and your proprioceptive system senses the applied and reactive forces your body experiences.
On a treadmill, however, you’re not really going anywhere (or at least not very far). That means you do not have to apply horizontal forces to overcome your body’s inertia to accelerate (with respect to the treadmill’s surface). An ODT simulates what it would be like to remain under the influence of gravity, and retain angular inertia, but eliminate linear inertia. This unusual condition is at odds with a person’s natural reflexes for directing their body’s motion, rendering their natural responses ineffective, i.e., making it difficult to stop, accelerate, or redirect one’s motion quickly on a treadmill.
ODTs are challenging to build because of the demands of their mechanical and control systems. I suspect that a lot of people believe that it is just a matter of perfecting ODTs to keep the user nearly centered of the treadmill in order to provide a realistic simulator for natural human locomotion. That is not the case.
There is a way to simulate linear inertia: Strap the user to an active mechanical tether that pulls and pushes on them to apply the forces necessary to substitute for the (simulated) absence of inertia. The Sarcos Treadport system did just that, albeit it is a linear treadmill:
http://www.cs.utah.edu/research/areas/ve/LocomotionDisplay.html
I have yet to see such a harness built to apply omnidirectional (horizontal) forces for simulating inertia on an ODT.
There may be another alternative. People can learn a wide variety of motion skills. They can learn to skate and ride a surfboard. With enough practice people may be able to master virtually accelerating and decelerating on a well-designed ODT. But they will not be performing natural locomotion and there is no reason to believe that their performance (speed & accuracy) at varied motion tasks will match their real world performance.
-JimT
25 Replies
- G'day Jim,
I've stewed on this one a lot myself and also see it as a major shortfall in current devices. What i'm uncertain of is which of the cues take precendence in our perception of acceleration (in any direction).
I'm working on a unpowered locomotion surface that would at least generate a form of angular inertia, such that to either accelerate/decelerate or initiate a turn would require more effort than just sliding on the spot for example. In this case, your legs and body would be giving feedback.
What is still missing would be the vestibular feedback you get from your head accelerating and decelerating in different directions. I'm sure i'll find out how critical that is when I finish my prototype and give it a run. In any case I think it will be an improvement on sliding.
Detlev - Who cares? Any ODT that frees up your hands is better than having to use a keyboard or a thumbstick. Things will improve over time.
I see a lot of splitting of hairs about things like ODTs on this forum and the oculus subreddit, and it really doesn't make any sense. So they aren't a holodeck locomotion solution. Big freaking deal. The Rift is hardly perfect either. Regardless, the current ODTs are *far* better than the alternative of using a thumbstick or WASD.
It's like complaining that your delicious chocolate cake doesn't have a cherry on top, and that you are therefore going to stick to eating dirt. "n00854180t" wrote:
Who cares?
What motivated me to post was seeing developers of ‘stepping-in-place’ systems apologize for their UI’s not being on par with ODTs. – That remains to be seen.
I’m posting on this forum to communicate with the current generation of designers & developers of VR interfaces. We’re participating in a collective search through a very large design space to come up with effective avatar controls."n00854180t" wrote:
Any ODT that frees up your hands is better than having to use a keyboard or a thumbstick.
There are ‘hundreds’ of different ways of achieving hands-free virtual locomotion control, most of which require too much conscious effort & fail to allow you to make your avatar perform in a natural way. I hope that I’ve at least dispelled the myth that a system that ‘simply’ keeps a person centered on an ODT will provide a truly natural locomotion simulator.
What do you really want to be able to do with a virtual locomotion control? Touring an architectural model is easy. Playing mobility-oriented sports is hard. Moving tactically in a combat simulator is really hard."n00854180t" wrote:
Things will improve over time.
All you need to perfect virtual locomotion on an ODT is an artificial gravity generator. Barring that we need to explore the realm of the possible."n00854180t" wrote:
I see a lot of splitting of hairs about things like ODTs on this forum and ...
It's like complaining that your delicious chocolate cake doesn't have a cherry on top, and that you are therefore going to stick to eating dirt.
But I consider myself & others to be Chefs, seeking to improve the state of the art. I used to think that conventional gamepad interfaces were terrible; now I appreciate how much they do with so little. As an end user, you’re lucky that so many poor UI’s are culled before you ever have to experience them.
-JimT"Simnertia" wrote:
What i'm uncertain of is which of the cues take precendence in our perception of acceleration (in any direction).
For cerebral matters that involve the neocortex, vision usually predominates. But balance & locomotion control also involves the spinal cord, brain stem, & cerebellum, so all bets are off."Simnertia" wrote:
I'm working on a unpowered locomotion surface that would at least generate a form of angular inertia, such that to either accelerate/decelerate or initiate a turn would require more effort than just sliding on the spot for example. In this case, your legs and body would be giving feedback.
It’s hard to give you feedback without having a better sense of your design.
I suspect that you’re providing enough traction to allow the user to move the surface (tread?) underfoot. To do so the user will need something to push his body against to keep from simply walking away. That suggests the user is either standing in a rig or wearing a ‘turning harness’. Once you have a rig or a harness the dynamics are different enough from free stepping that you don’t need to worry about your reflexes working against you (so much).
I like the idea of having some traction rather than ‘just’ a slippery floor, but can you make it feel better than a stationary floor with just the right amount of kinetic friction? There are also pros & cons between traversing the support foot versus simply walking-in-place.
Rings & ‘turning harnesses’ have there own set of design issues. I worked with Professor Roger Kaufman at GWU to come up with a ‘Go-prone turning harness’. See: http://www.seas.gwu.edu/~kaufman1/NRL/Gaiter.html
This remains a research topic unto itself."Simnertia" wrote:
What is still missing would be the vestibular feedback you get from your head accelerating and decelerating in different directions. I'm sure i'll find out how critical that is when I finish my prototype and give it a run. In any case I think it will be an improvement on sliding.
Detlev
With a ring or turning harness I don’t think you have to worry too much about the user loosing his footing due to vestibular imbalance. It’s a bit more like pushing (or pulling) a shopping cart around. I’m glad to see exploration in this area. It’s a challenging mechanical engineering design problem. I wish you the best & hope to hear how it turns out.
-JimT- You are going to need dozens of muscles to accelerate and deccelerate on an active ODT as well as for keeping the balance. So at least part of proprioceptive afferent nerves in muscles, joints will send information about that to the brain. MAybe your vestibular system doesnt respond but part of proprioception would still work.
An active ODT wont simulate natural working 100% but already 90% would be a great outstanding achievement
until nanobots start to appear in 15 - 1000years.
Passive ODts wont simulate natural walking at all, but an Infinadeck ODT can come really near.
Best way would be to guide some cables into our ear to our accelerometer the Sacculus and Utriculus
and manipulate it right there. But how do you overcome the eardrum? :mrgreen:
Actually it is a surgical practice to pierce sometimes the drum and leave even sometimes a T-tube to prevent accumulation of unwanted liquids. http://en.wikipedia.org/wiki/Tympanostomy_tube
Actually its quiet big with 1,5mm and can stay for 12 months in the drum. The body rejects it afterwards and the drum regenerates.
So no that far fetched. :lol: "JimT" wrote:
"n00854180t" wrote:
Who cares?
What motivated me to post was seeing developers of ‘stepping-in-place’ systems apologize for their UI’s not being on par with ODTs. – That remains to be seen.
I’m posting on this forum to communicate with the current generation of designers & developers of VR interfaces. We’re participating in a collective search through a very large design space to come up with effective avatar controls."n00854180t" wrote:
Any ODT that frees up your hands is better than having to use a keyboard or a thumbstick.
There are ‘hundreds’ of different ways of achieving hands-free virtual locomotion control, most of which require too much conscious effort & fail to allow you to make your avatar perform in a natural way. I hope that I’ve at least dispelled the myth that a system that ‘simply’ keeps a person centered on an ODT will provide a truly natural locomotion simulator.
What do you really want to be able to do with a virtual locomotion control? Touring an architectural model is easy. Playing mobility-oriented sports is hard. Moving tactically in a combat simulator is really hard."n00854180t" wrote:
Things will improve over time.
All you need to perfect virtual locomotion on an ODT is an artificial gravity generator. Barring that we need to explore the realm of the possible."n00854180t" wrote:
I see a lot of splitting of hairs about things like ODTs on this forum and ...
It's like complaining that your delicious chocolate cake doesn't have a cherry on top, and that you are therefore going to stick to eating dirt.
But I consider myself & others to be Chefs, seeking to improve the state of the art. I used to think that conventional gamepad interfaces were terrible; now I appreciate how much they do with so little. As an end user, you’re lucky that so many poor UI’s are culled before you ever have to experience them.
-JimT
Regardless, thumbsticks/WASD are still crap compared to an ODT or really anything that frees up your hands from handling locomotion.
So complaining that the literally first ever commercial ODTs are not perfect holodeck locomotion simulators seems pretty ridiculous to me.
If you're a chef, don't complain about the filet mignon of locomotion inputs (ODTs) in order to go eat mud (thumbsticks/WASD).
I don't see the point. It's not productive to rehash these complaints over and over. If you truly believe that thumbsticks/WASD are the one true locomotion solution for all time why are you even talking about ODTs? Go use a thumbstick and be happy.
It's the same as people who complain the Rift isn't literally a holodeck or the Matrix. It's absurd.
Not to mention how hyper obvious a complaint like "The first ever commercial ODTs are not LITERALLY THE MATRIx". It's like "No s**t Sherlock!""cerebral" wrote:
You are going to need dozens of muscles to accelerate and deccelerate on an active ODT as well as for keeping the balance. So at least part of proprioceptive afferent nerves in muscles, joints will send information about that to the brain.
You certainly feel the reactive force of it holding you up against gravity.
Walking on a treadmill captures the kinematic aspects of locomotion, but not the kinetics:
Your legs are going through the motions, but your body isn’t accelerating through space.
Clearly there are forces involved with moving the legs forward and back faster.
When your legs are at an angle to the floor they convey lateral forces to support your body."cerebral" wrote:
MAybe your vestibular system doesnt respond but part of proprioception would still work.
But the one thing that’s missing from this equation are the forces applied to accelerate & redirect your motion. That alone is enough throw off your natural abilities to quickly stop or change your direction of motion.
This paper specifically addresses having to compensate for not having to overcome inertia:
Christensen, R. R., Hollerbach, J. M., Xu, Y., & Meek, S. G. (2000). “Inertial-force feedback for the treadport locomotion interface.” Presence: Teleoperators and Virtual Environments, 9(1), 1-14."cerebral" wrote:
An active ODT wont simulate natural working 100% but already 90% would be a great outstanding achievement
I’m not saying that people cannot learn to ‘apply the brakes’ in a different way on an ODT than they do in the real world. Maybe that will be considered ‘fun’ or acceptable. On the other hand is such an artificial UI better than ‘simply’ stepping-in-place or other techniques for the motion tasks they want to accomplish?"cerebral" wrote:
Best way would be to guide some cables into our ear to our accelerometer the Sacculus and Utriculus
and manipulate it right there. But how do you overcome the eardrum? :mrgreen:
I favor non-invasive approaches:
Riccio, G. E., McMillan, G. R., Cress, J. D., & Schnurer, J. H. (1995). Functional Transcutaneous Electrical Stimulation of the Vestibular System. Illinois Univ at Urbana.
-JimTOn the other hand is such an artificial UI better than ‘simply’ stepping-in-place or other techniques for the motion tasks they want to accomplish?
I am highly confident that an active ODT like the Infinadeck will be much better than the stepping in place techniques.
But i understand your points.
Can transcutaneous galvanic vestibular stimulation really give you the feel of acceleration or just precisely disturb your head orientation as if you have drunk too much?
Talking about invasiveness. Isn't galvanic vesibular stimulation in itself a bit crazy? ;)
The vestibular system would be actually quiet reachable as a first 'nerve interface'. A 50 micron cable in your ear penetrating the ear drum wouldnt be that mad i think. It would be also possible to reach the middle ear/vestibular system via the nose. Maybe a good application for a bigger microbot(100-200µm).
Would it be possible to stimulate the vestibular macula organs with ultrasound since the cells are mechanoreceptors?
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3960553/- I've never understood why people simply dismiss natural walking. Sure there is an issue of practicality but faced with these problem checklists:
ODT:
- Build large, complex, expensive, treadmill mechanism that enables unencumbered motion and sudden velocity changes
- Build tracking solution to predict and counteract player direction of motion
- Build invasive device to simulate inertial effects
Redirected walking
- Find empty space approx (100' x 100')
- Build tracking solution (IMU suit, optical camera array, etc.)
It's readily apparent that natural walking is vastly simpler, cheaper, more versatile, more practical, and more realistic. Why would anyone endeavor to tackle dangerous and probably unsolvable problems like nerve and vestibular stimulation instead of the totally brain-dead easy problem of finding an empty warehouse? There's really no reason to wait for all these unsolved engineering problems to come together to enable stationary locomotion. Motion stages are a proven solution and available with today's technology. - Buying a $1000 ODT versus renting (min. lease is usually at least a couple of years) a several thousand dollar per month (not including utilities like water or power) space.
It's not hard to understand why people aren't super keen on the latter. Plenty of enthusiasts can afford a one-time expense of $1k.
The same cannot be said about a monthly expense of $2k+ not including utilities, which also usually requires a down payment of multiple months' worth of rent, etc etc.
I think you've seriously jumped the shark if you're actually trying to argue that it's more convenient to rent a 100'x100' space than it is pay half of one month's rent on such a place towards a one-time expense on an ODT.
It's about the same as claiming that it's more practical to literally build a new universe than it is to buy a Rift, and just as ridiculous.
