Up and Coming

Centering Mechanisms for Stepping-in-Place Locomotion

6 replies
2555 views

Underlying Issue: When a person walks-in-place for any length of time wearing a closed bezel HMD (or with their eyes closed) they tend to drift off center. If the user is not re-centered he may move out of tracking range, or bump or trip on a physical obstruction.

I will confine this discussion to walking-in-place virtual locomotion controls because other techniques, like sliding the feet against the floor, impose their own set of requirements.

General Criteria
Ideally we want something that does not introduce sensory or cognitive distractions, or disrupts the user from what he’s doing in VR, while constantly keeping the user from moving off center. Distractions include having to pay attention to additional sensory cues and simply having to remain aware of where you are in the real world. Having to perform adjustments on an ongoing basis that you would not do in the real world is a background disruption. A foreground disruption is an interruption: having to stop what you’re doing and perform an adjustment (having to take a ‘time out’). If you can ignore the centering technique when focusing on immersive task, and if ignoring it produces dire consequences (like waking into a wall) then it is inadequate for its purpose.

Physical Centering Systems
Turning Harness: It provides a springy force to keep you centered while allowing you to turn freely. It pushes back against the drifting that would otherwise occur during stepping-in-place. It must have some lateral & vertical give and allow your hips to sway; otherwise you can’t step-in-place in a natural manner. It serves to push you back to the center, gently at first with the force increasing as you move further away.
Pros:
- You are always in contact with it, so it provides constant directional haptic feedback.
- Once you get use to stepping-in-place with well designed harness it neither distracts or disrupts your virtual locomotion.
- Leaning into a harness when you run-in-place makes it feel more like natural running.
Cons:
- It requires a somewhat complex mechanism & mounting scheme. – Added cost.
- It must be donned & adjusted to fit.
- Although go-prone turning harnesses exist they have yet to be perfected.

Standing in a Ring: It’s usually a fixed ring, with about a 1m radius.
Pros: - Cheap, if kept simple. - Provides obvious directional haptic feedback.
Cons: - You may bump into it abruptly, causing a minor distraction.
- Obstructs the arms & lower body postures.

An ODT: It would be used to re-center the user. Since rapid motions are performed by stepping-in-place, it only needs to perform low speed translation.
How easy is it to step-in-place with the ground slowly moving beneath your feet? I have not had the opportunity to try stepping-in-place on an ODT, so I’m not sure how distracting or disruptive it might be to have the tread moving under foot. If this not a problem then this would provide a good alternative.
Pros: - Just walk onto it to use it. - Very little interference with other natural movements.
Cons: - It has many moving parts. - It needs to be sturdy enough to jog in-place on.

Cue-Based Centering Systems (these could be combined)
All cue-based centering techniques impose both distractions and disruptions. The cue is a distraction. Since the user must react to the cue it presents at least a background disruption. Since cue-based techniques cannot physically guarantee the user doesn’t remain centered the user might find getting back on track to be an interruption.

Standing on a Textured or Slightly Raised Surface
Pros: - Very simple & cheap. - It does little to mess up your living room.
Cons: - It provides rather sparse directional feedback: ‘I know one foot is off, but where’s center?’

Standing in a Dish or on a Mound
Pros: - Fairly simple & cheap. - It provides better directional feedback than a flat mat.
Cons: - At what point does its curvature interfere with stepping or pose a trip hazard?

Haptic Feedback: Wearing a belt with tactors arranged at 45 degree intervals to provide directional feedback w.r.t. going off center. – I’d prefer to reserve this to indicate making contact with virtual surfaces. Mild electrical shocks could be applied in the same way.

Visual Feedback: There are too many different options to fully cover here. It needs to be directional & non obtrusive. It can become obtrusive if the user moves too far off center, but having to do so is an admission of failure.
Pros: - Free with every HMD.
Cons: - Adds visual clutter/distractions to the user’s view.

Open Bezel HMDs: a special case of providing visual feedback to allow the user to re-center himself.
Pros: - Its simple & cheap. - Some people claim it helps reduce sim-sickness.
Cons: - The user is now responsible for remaining aware of where he is in both the virtual and real world. Attending to the virtual world in your central vision and the physical world in part of your peripheral vision sets up some perceptual discrepancies.

Audio Feedback: It’s tricky to make this directional without interfering with other audio cues.

Sweet Spots
Having a moderate sized sweet-spot – say 2m x 2m: You could set up a limited space in which the user can move around ‘freely’ before the centering condition kicks in. The problem with this is that the user has to deal with two different modes of operation in terms of moving within the space and at the boundary. And without sufficient warning the user may not anticipate making contact with the boundary.

Having a fairly large sweet spot – say 4m x 4m: Then the user can choose to walk naturally for short distances or step-in-place to cover long distances. A problem with this is that the user needs to be able to manually re-center himself to make the most of the walking space. Thus the user needs to tell the system when he’s intentionally moving back to re-center himself by engaging a ‘clutch’ to avoid making virtual movements. – This means taking a time-out, which can be disruptive.

Conclusion
The advantage of the Turning Harness is that it constantly keeps the user centered even when it is ignored. This minimizes the distractive & disruptive effects involved in re-centering yourself, and thus enhances the virtual locomotion experience.

-JimT

mptp

Up and Coming

This is a great post. I'll put aside my issues with walking-in-place for locomotion and say what I think:

I think a physical centering system would defeat the purpose of the exercise. Walking in place is an attractive solution because it's cheap, intuitive and unobtrusive. Having to strap in to something that costs several hundred dollars is silly when you could instead buy an Omni/Virtualizer for only a bit more and have an overall better experience. Of course, they have their own issues...

If the only thing you're worried about is people slowly drifting off centre, you could employ a similar technique used on highways to stop cars slowly drifting off the road. Have a surface with a ring of bumps around the outside of a 1x1m 'sweet spot', so that when someone is drifting off the sweet spot, they can feel the bumps and know to take a step in the other direction. I would think this would only happen after quite a bit of walking-in-place, so the loss of immersion wouldn't be something to worry about.

Melbourne-based creative technologist. I flit between experimental AR/VR experiences, audiovisual electronics and full-stack web development. http://www.lachansleight.io

JimTAuthor

Up and Coming

"mptp" wrote:
This is a great post. I'll put aside my issues with walking-in-place for locomotion and say what I think:

Thank you.

"mptp" wrote:
I think a physical centering system would defeat the purpose of the exercise. Walking in place is an attractive solution because it's cheap, intuitive and unobtrusive. Having to strap in to something that costs several hundred dollars is silly when you could instead buy an Omni/Virtualizer for only a bit more and have an overall better experience. Of course, they have their own issues...

Yes, a good turning harness for either walking-in-place or sliding-in-place is going to be a non-trivial piece of equipment. It’s not clear how a well designed walking-in-place simulator compares to well designed a sliding-in-place simulator. Too many of the walking-in-place systems that I’ve seen are rather crude, converting footfalls to virtual steps in the direction some part of the user’s body is facing. A good walking-in-place system should continuously tie the extent and direction of leg movement to the avatar’s virtual displacement.

Never judge an immersive user interface by how it looks from the outside!
Does sliding-in-place feel more natural than walking-in-place? Does it make it easier & more graceful to turn your body while continuing along a path? How does it coordinate with other actions, like crouching, looking, reaching, & aiming?

A strength of walking in place is that its compatible with actual walking: you can stop moving & run-in-place (if the floor is not too slippery). Admittedly, this advantage is negated by wearing a turning harness. Stepping-in-place can be combined with redirected walking, but there’s still the issue of the user’s free will: he may choose to actually walk in any direction at any time.

"mptp" wrote:
If the only thing you're worried about is people slowly drifting off centre, you could employ a similar technique used on highways to stop cars slowly drifting off the road. Have a surface with a ring of bumps around the outside of a 1x1m 'sweet spot', so that when someone is drifting off the sweet spot, they can feel the bumps and know to take a step in the other direction. I would think this would only happen after quite a bit of walking-in-place, so the loss of immersion wouldn't be something to worry about.

It doesn’t take long to drift off center when vigorously walking-in-place (much less running-in-place). If you have to ‘watch your step’ to carefully avoid drifting, the mental burden is already too high.

Speed bumps on a road simply let you know when you’ve gone too far to one side. Centering yourself back into a circle requires more directional feedback. It’s hard to deduce where you need to go by feeling a single bump with your foot. Adding visual cues in the HMD gives more directional feedback at the cost of greater distraction.

The main point I’d like to make is what “the loss of immersion” really means. If you simply want to experience a new form of virtual locomotion, then having to stop & reorient yourself isn’t a big deal. But if you participate in a game or social interaction involving virtual locomotion, such interruptions can undermine your performance to the point where you will reject the control technique. It can be down right embarrassing to have to fumble back into place when a moment before you were moving with intention through the virtual world, and even a slight pause can throw off your timing.

The conclusion that I’ve reached, after more than 15 years in the field, is that the goal is ‘behavioral realism’: to match the performance of the user’s avatar in VE to the user’s performance in the real world. (If they have to don a HMD & harness or sit in a chair & apply foot pedals, then so be it.)

-JimT

mptp

Up and Coming

"JimT" wrote:
The conclusion that I’ve reached, after more than 15 years in the field, is that the goal is ‘behavioral realism’: to match the performance of the user’s avatar in VE to the user’s performance in the real world. (If they have to don a HMD & harness or sit in a chair & apply foot pedals, then so be it.)

Hmm, the problem is the goal isn't necessarily behavioural realism at the cost of all else - I feel as though it's 'as much behavioural realism for as many people as possible'.
Meaning, if it's expensive or bulky, less people will buy it, so it doesn't matter if it's the best system of all time.

But honestly, how I feel about it is if you're going for behavioural realism, then you should stop thinking about stepping-in-place locomotion in the first place. It's better than a joystick, but little better than a lean-input, with the added downside of being physically tiring.
If you're going to have the user physically do walking motions, then you should aim to make those walking motions as close to the actual pendulum-like motion of the legs during normal walking/running.

...how you achieve that is a whole different issue. :P

Melbourne-based creative technologist. I flit between experimental AR/VR experiences, audiovisual electronics and full-stack web development. http://www.lachansleight.io

JimTAuthor

Up and Coming

"mptp" wrote:
Hmm, the problem is the goal isn't necessarily behavioural realism at the cost of all else - I feel as though it's 'as much behavioural realism for as many people as possible'.
Meaning, if it's expensive or bulky, less people will buy it, so it doesn't matter if it's the best system of all time.

Yes! Look carefully at my definition of ‘behavioral realism’ and you will see that it opens the door to substituting less expensive solutions, so long as they provide excellent control over your avatar.

Contrast:
- Behavioral Realism: to match the actions the user’s avatar performs in VE to the physical actions the user performs in the real world.
with
- Physical Realism: to match the physical actions the user performs when interacting with VE to the physical actions the user performs in the real world.

Physical realism is required to train physical skills, but behavioral realism is sufficient for training cognitive skills (e.g., tactics). For a virtual locomotion control to provide behavioral realism in a ‘FPS’ it needs to (1) provide a full range of movement capabilities: walk, run, crawl, vary postural height (crouch & high-low prone), and allow the user to independently vary the heading (turn the body) and course (direction of motion w.r.t. the body), and (2) coordinate realistically with the other actions involved in ‘FPS’: looking, shooting, and taking cover.

To date, the best system I’ve seen that supports moving to & using cover effectively (other than actual walking) is Pointman: http://en.wikipedia.org/wiki/Pointman_(user_interface). It achieves a high level of behavioral realism, but little physical realism. Is that a legitimate tradeoff?

So far, most of the work in VR has focused on providing realistic sensory input, with immersion tied to surrounding the user with perceptually realistic ‘imagery’. Now we’re seeing more exploration of motor control over the user’s avatar. I posit that seamlessly realizing your intended actions in VR with behavioral realism will in general, prove more important than physical realism, but that remains to be seen.

An irony is that many people envision either total physical realism or total brain control as the ultimate VR UI, without considering a hybrid, or whether or not a brain controlled interface would be useful for training physical skills.

"mptp" wrote:
But honestly, how I feel about it is if you're going for behavioural realism, then you should stop thinking about stepping-in-place locomotion in the first place. It's better than a joystick, but little better than a lean-input, with the added downside of being physically tiring.

A problem with a lean-input (for controlling the course) is the way it interacts with turning the body. In actual waking you step-to-turn which coordinates well with stepping-to-translate, the same is true for stepping-in-place; but it’s awkward to maintain your direction of lean while turning your body in place. And what happens when you want to lean your avatar to look around a corner?

A key question is: Do we want the virtual locomotion technique to be as tiring as natural locomotion? If you want it to be a realistic simulation, then yes; but if you want to give up some realism to do more ‘runs’ than you can physically perform, then no. A similar argument can be made for standing versus seated controls.

"mptp" wrote:
If you're going to have the user physically do walking motions, then you should aim to make those walking motions as close to the actual pendulum-like motion of the legs during normal walking/running.

The pendulum-like motion of the legs is but one of the many characteristics of natural gait. Both walking-in-place & sliding-in-place look promising, so long as an effective centering mechanism is provided.

-JimT

Techfree

New Participant

Hi,

I will expose a concept for your project and give me feedback.

http://imageshack.com/a/img908/3/WGUj1Q.jpg

http://imageshack.com/a/img661/4204/2trkow.jpg

The system consists of a circular mobile platform covered with a low friction surface, in this platform have set Omnidirectional wheels.
The center of the platform is provided with a circular base which vien be clamped in the fixed base between lesquelle we attach a spring. This could also be a cable winder system has automatic return so that the platform back to the center.
Fixed base is overcoming a mat just hold the hip via a mobile horizontal belt automatic return.

Regards Mohamed.

JimTAuthor

Up and Coming

"Techfree" wrote:

I will expose a concept for your project and give me feedback.

I don’t fully understand how your proposed system is intended to work:

> Is the base intended to turn, so that the user doesn’t physically turn (very far)?
--- If the user physically turns then the hip-belt mechanism must turn with them, adding complexity.

> Your figure showing a shoe shows high-friction material under the middle of the curved sole, flanked by low friction material. The parts at the front & back ends of the sole are unlabeled. The figure showing a man walking in different phases of gait seems to indicate that the front & back ends of the sole are areas of high friction.
--- It has been my experience looking at the way different people walk and run in different directions and step to turn and pivot, and combine these actions in different ways, it becomes difficult to consistently characterize the way heel & toe strikes occur and the way the center of pressure shifts as a step is taken.

-JimT

Get Help

Sign up

Login with SSO

Login to the community

Login with SSO