Great Strides - The VR Locomotion Problem (First Post)

Talk about Head Mounted Displays (HMDs), augmented reality, wearable computing, controller hardware, haptic feedback, motion tracking, and related topics here!
Post Reply
MattStompz
One Eyed Hopeful
Posts: 3
Joined: Sun Apr 13, 2014 7:02 pm

Great Strides - The VR Locomotion Problem (First Post)

Post by MattStompz »

Great Strides - VR Locomotion

I have seen quite a bit of interest in Virtual Reality locomotion in the past few months. As VR starts to come into its own and prove to be a viable industry, a lot of attention is focused on the next logical problem; input. Hand tracking is the first input hurdle, and there are a lot of brilliant methods for providing true 1:1 hand capture, including; the VIVE/HTC Lighthouse marker-less control system, Sixense STEM’s absolute magnetometer/IMU tracking, Sony’s Move controllers, and true LIDAR/time-of-flight hand tracking using Leap Motion or (Oculus’) Nimble Sense.

But tracking hands for VR input, while technically challenging, at least has an easily recognized solution; good hand tracking will precisely input your hand’s exact position and orientation in VR. So why doesn’t this same solution apply to VR’s second big hurdle; locomotion?

The “Stepping Stones” problem

Imagine you’re standing on the bank of a river (in VR, of course) and there are several large rocks positioned at different distances that, should you leap from one stone to the next, will allow you to cross the river. The problem is that you are not on the bank of the river, you are standing in your office and you have some fairly expensive equipment you’d rather not body slam, and a desk with some nasty corners that like to remind you that you have terrible night vision. How do you precisely cross this path in VR without damage to life and limb in reality? It’s a significant issue. You need the inertia from each leap to translate into the game, but absolute positional tracking doesn’t work for this because it tracks the position absolutely, and we’re trying to keep you from jumping into your desk.

To allow you to cross our VR river, we need to interpret your movements from a static position whether you are seated, standing, supported in a constraining harness on a slidemill, or on a true, mechanical, omni-direction treadmill. This means recognizing the amount of force with which you are pushing off and measuring the trajectory of that intended leap, all without letting you actually move. Some potential solutions to this are to interpret the desired inertia of the push-off by tracking the speed and vector of the foot as the user either:

A) Jumps-in-place without forward movement
B) Slides their feet quickly on a slidemill in a sort of quasi-jump
C) Uses a true treadmill with an advanced control scheme to recognize and negate the movement with mechanical inertia (i.e. comfortably and carefully keeping the user at the center of the walking area).

But how do we know 100% that the user’s intention was to leap? How do we know they weren’t trying to do something else?

This brings us to our next big issue.

The “Happy Feet” problem

So now you’re on the other side of the river and it’s time for a celebration. You’re going to do your famous end-zone dance and break it down for the virtual ladies. But, instead of your flawless interpretation of a sprinkler, you have accidently leaped back into the VR river and are now being swept away.

What happened?

That movement that you had intended to be a one-legged victory dance was interpreted as the same movement meant to let you leap safely to the next stepping stone in our last example.

There are times when users will want to use absolute positional tracking to do jumping-jacks, juggle a virtual hacky-sack, or just get jiggy with it. How can we determine what the user is attempting to do? Again, it is not a simple problem.
One potential solution is to have an interpreted movement mode and an absolute position mode that is toggle-able by the user. But then, how do you play a game of soccer where the user needs to dribble a ball while moving? Another solution is to allow joystick/gamepad movement for locomotion and absolute positional tracking for everything else (although this doesn’t really address our problem).

Absolute positional tracking and locomotion don’t really mix well unless you have a very large empty space and some way to let you walk through it with 1:1 accuracy.

But even then there are issues.

The “Dodge, Dip, Duck, Dive, and Dodge” problem

So, we’ve got our absolute positional tracking working flawlessly with the locomotion enabled and now we want to play our game. We’re running through the battlefield, dodging bullets and ducking debris. We’ve managed to flank the enemy by dropping into a ditch and dipping through a hole in the fence-line. Now you need to sneak up on the enemy turret, so you dive to the grass and start to low crawl up the hill. And, although you haven’t taken a single hit in-game, you might be suffering from a bruised rib, a sprained wrist and a slight concussion in reality.

This is because, even if you have a large, dedicated room without furniture and a wireless headset that warns you before you could run into the walls, unless you have a perfectly level terrain through which you are maneuvering, the environment you play in, and the actual room around you, will not match. Of course these movements also require you to be unrestricted, and the safety of the constraining harness is now also limiting the actions you can complete. Not having the sensation of actual changes in elevation as you climb and descend is also making you unbalanced and nauseous.

Solution: Make all VR games completely flat.

If you are satisfied with that answer then that’s that. If you’re not, then you see that this is a very difficult issue because we, essentially, need to simulate various terrains in reality. This problem is similar to haptic feedback because it will mean adjusting the elevation of the actual ground the user is standing on (i.e. a platform) to match the in-game elevation; not a simple feat.

The problem with “Perfect” locomotion

So what will it take to give you the perfect VR locomotion experience?

You’ll need a moving “carpet” that is a true, 360 degree, mechanical omni-directional treadmill which uses perfect, absolute positional tracking to keep you at center no matter what you do. You can dance and shake in the center until you can shake no more and then take off running at a full sprint without even budging in reality. This treadmill does not need any kind of support or constraining structure so, if you can manage it, you can do it. If you can do a backflip, it will let you backflip (we’ll have completely wireless headsets by this point, I hope). But don’t worry too much, because the moving “carpet” will be soft enough to cushion a fall. This treadmill also adjusts to any terrain type (within reason) and stymies you with haptic feedback when you run into walls in game.

Sounds pretty sweet, right? So where’s the problem?

Cost is the greatest issue. This is not a consumer product. Even when VR becomes so main-stream that it is in every house, this will probably be the equivalent of building a full gym and big-screen theater in your house (of course you won’t have need for either if you have this, will you?).

The technology just isn’t there yet. We are working on getting it there, but it is still a long way off (if you’d like to see who is closest to this full solution search for “Infinideck”).

Our solution

We have been researching VR locomotion for a long time, and we have sought the perfect balance of low-cost and high-tech. Keeping the input affordable and accessible has always been our primary focus. The StompzVR device enables realistic movement without any restricting constraints to let you take full advantage of the new positional tracking solutions that are coming out. We are developing creative solutions that let you dance when you want to and then take off running when you are ready (I hope you appreciate our candor so far, and understand that we can’t just give everything away). We are actively involved with the VR community to try and solve all of these new, exciting problems.
Our goal is to provide the perfect VR movement experience. We made StompzVR so you can use it anywhere and with any game, without the need for tons of setup or software integration. The perfect future VR locomotion system might not look like StompzVR or anything else that is out there right now, but we have been working on the problems with VR locomotion for a long time and we feel we’ve created the best solution to bring VR movement to the masses, now.

Thanks for reading,
Matt "Stompz" Carrell
http://www.stompzvr.com
------
Post Reply

Return to “General VR/AR Discussion”