Whether you’re an old VR hand or one of the legions of newbies discovering the technology during this time of lockdown and tele-whatevering, you’ve probably discovered a couple of things about VR.
One is: VR is just plain awesome. There is nothing in our previous computing experience like being able to look around freely and naturally, have full depth perception, and “touch” and manipulate game objects with our hands (either directly or via controllers). It’s a whole new world, and it puts a big dumb grin on most of our faces. [I think affordable VR is one of those historical inflection points in UI technology — like the first windowing OS, or the first touchscreen pocket device — when the world changes; but that’s a topic for another essay.]
The next thing you notice — if you’re me anyway — is that your avatar in most VR sims and games has no legs! The amazing realism of VR environments — the incredible immersion compared to “pancake” (flat-screen) gaming — stops at the hips (where your hands come to rest). You have no legs. There are no sensors on your feet. You can’t walk, kick a ball, or dance.
Well, actually, in many VR worlds you can walk around, even if your feet aren’t being tracked. Relative motion is well-scaled, head movement is perfectly natural, and it feels great — incredibly immersive. But there’s one serious gotcha: most of us have playspaces small enough that you can’t walk very far before you hit a boundary. And even those few of us who can afford to dedicate a sizable room to our VR habit still can’t walk through Half Life: Alyx! The bigger and more challenging the virtual world, the more the playspace boundaries constrain our ability to explore it.
In our VR dreams, we walk freely as far as we like through magical worlds of adventure, exploration, and virtual tourism. But in our present VR reality, we run into the furniture.
The game devs have thought about this a lot and come to a convergence, if not a perfect consensus; if you want to get around in the VR world, there are two industry-standard solutions. One is to use one of your controller mini-joysticks (usually the non-dominant hand) to “glide” or zoom around — kind of like driving a drone, except the drone is you-in-VR. The other is to use one of the buttons on a controller to throw an illuminated target spot out in front of you, then teleport to that target location.
So gliding and teleportation are the prevailing methods of locomotion in VR worlds today. And that solves the problem of limited playspace. While standing or sitting still inside your boundary, you can explore expansive worlds.
Sadly, both of these methods are disappointing. Some people feel a little (sometimes a lot) motion-sick, when skittering around like an air hockey puck while their physical body remains at rest; and teleportation is just so darned immersion-breaking, like it reminds you every few seconds that you’re in a game/sim. Neither method engages your body realistically, in the way that VR otherwise engages your hands and arms to grab, throw, climb, fight, swim and so on, plus your head and neck to look all around you. Tracking head position allows you to bend, crouch, and stand on tiptoe realistically… but you can’t walk or run.
Some folks resort to walking or running in place while using the joysticks to glide around; this can reduce the puke factor quite a bit. It provides the natural head-bobbing view of your surroundings that says “I am walking” to your brain, as well as the familiar rhythm of stepping with feet and legs. But really, you still have to use a joystick to move and steer; and you’re trying to match your pace to the gliding motion, instead of moving normally at a speed determined by your pace. Natural-feeling locomotion in VR just isn’t there yet.
At the fringes of every industry are people trying to solve the hard problems by unconventional means. The VR locomotion problem is no exception. In one camp there are people working on captive running rigs: physical devices that let you run and walk freely while tethering your physical body in place. These rigs tend to be expensive and bulky, of course; but they’re darned ingenious.
The most exotic and (to me) droolworthy examples are the fabulous Omnideck and the only slightly less fabulous Infinadeck. These are large, complicated and definitely not-for-the-home-market gizmos that use active positional compensation to keep the player centered on an x-y treadmill surface no matter how fast s/he runs or walks. These are the VR floors of our dreams, that allow infinite locomotion in any direction. Run, walk, turn corners, no problem. But the price tag? Don’t even think about it. For enterprise clients only.
Further down the price chain we can find “skating in a bowl” slidemill variants like the Kat VR product line or the Virtuix Omni. But the thing about slidemills, as Packet39 blog ruefully admitted back in 2018, is that they don’t feel like walking…
Slidemills do one thing and one thing only. They conform to the naive notion of what a VR treadmill should look like. It looks just like walking but in reality it is nothing like walking. It uses a different set of muscles, the friction is different, your balance is different, it’s cumbersome, tiring, painful, un-fun and just overall terrible. Look closely at all the videos above and take note how people shuffle their legs around. It’s just as uncomfortable as it looks. On top of that, it’s really limiting. You can’t crouch (not really, not fully). You can’t bend over to pick up something from the floor. You can’t duck behind a corner.
Maybe these slidemill issues will get worked out, but for the moment what you’re really doing is cross country skiing inside the dished top of a steel drum… it can be a workout, but it’s not walking.
The folks behind StriderVR have a different idea: a modest sized platform that permits omnidirectional walking in VR, but not running, fast side stepping, or jumping. While this might be a fine solution for virtual tourism or mild fitness workouts, it seems a little underspecified for active gaming. It’s also rather narrow in its target market (requires a Kinect 2) and currently in prototype, not production.
Continuing down the line from “most to least costly” we find a half-body VR suit (lower body only) called WalkOVR which uses 5 strap-on sensors plus a SteamVR driver layer to add leg/foot movement into your gameplay. Dunno about you, but having to “get dressed” for several minutes before playtime sounds a bit intimidating, and those straps don’t look very comfortable. Maybe for hardcore gamers only; and in any case, it’s presently “out of stock.”
Lower still in the price chain are clever but sketchy hacks like rocker boards that let you pump your feet while sitting in a chair (hardly immersive), or Cybershoes that work similarly, requiring a special carpet/mat for its wheeled shoes to “skate” on. Or the VRgo gaming chair which lets you tilt and spin to move in VR. Basically, these gizmos allow you to “walk and run” … while sitting. Which I guess is better than nothing, but it’s not the immersion level that I dream of. I love being able to play standing up, and also getting exercise while gaming.
The other line of approach to the Legless Problem is software; there are various hacks, patches and demos of ways to use your body motion (or a couple of extra trackers for feet) to get the feeling of natural walking and running in VR. The appeal of the software solution is obvious: it doesn’t involve a large additional investment in bulky equipment, you’re not strapped or tethered to a frame, and you’re not forced to make unnatural motions that then get mapped into walking.
VRocker is a Steam game controller patch that uses head-bobbing motion to activate the standard twin-stick gliding locomotion mechanism. By bouncing gently in place, or rocking from foot to foot, or walking/running in place, you can move around. The patch is very tunable to your personal preferences; many people have found it enjoyable, especially with action/combat games involving run/crouch/stalk tactics. The demo is fairly convincing, too. But it doesn’t work with all games — and it needs tuning to create a custom profile for each game for each user.
Natural Locomotion is another Steam driver/patch that activates joystick navigation by means of arm-swinging or the addition of additional “lighthouse” trackers to your feet. Some gamers like arm-swinging as a locomotion method, but obviously you can’t be aiming your longbow while swinging your arms. Setting up additional traditional trackers and integrating them into your Steam gaming setup rather weakens the standalone, portable appeal of the Quest 2 with its inside-out tracking; and it’s complicated, not for the average user.
Freedom Locomotion VR by Huge Robot (George Kong) is a demo app, not a patch or driver for existing games; it’s the one I’ve spent most time with and I was so delighted with it that I’m going to reproduce here most of my Steam review:
FLVR is one of the most exciting proof of concept demos I’ve seen for VR. Like many players I’ve been frustrated by the “legless” quality of VR gaming. Much of the immersive potential is lost when upper body interactions are so natural, yet the lower body remains immobile and aphysical methods like teleportation and gliding are used to move around. I’ve said often that I wish I could walk/run in place for locomotion, and even thought about building a DIY USB controller hack for flatscreen gaming for just that purpose.
Then I discovered Freedom Locomotion’s CAOTS method, and it’s bloody brilliant. It’s exactly what I imagined, and it should be the industry standard (are you listening, industry?). It makes “walking” in-game far more immersive and solves the problem of limited roomscale, permitting infinite walking distance. It adds a physical activity/exercise component to adventure and exploration games as well as virtual tourism. Truly you have to try it to realise how radical, yet how very sensible and simple, it feels. To use it, you simply walk, jog, or run in place.
Tuning the parameters takes a little time. You’ll need to do some iterative tweaking of the walking speed (and thresholds between walk, jog, and run) so that it approximates your intuitive expectations for your own pace. It can also be a little challenging to stay truly in place and not drift while jogging or running; but the demo’s virtual fence plus your Guardian perimeter will keep you from colliding with the furniture :-) It took me about 15–20 minutes of play time and tweaking to get comfortable with CAOTS — and then I loved it, like “where has this been all my VR life?”
Immersion was unprecedented; at times I truly felt I was in the virtual environment. I didn’t feel any of the vertigo that “magic carpet” gliding induces; I felt solidly anchored to the game world.
In my personal best of all possible worlds, the CAOTS algorithm is adopted by all headset manufacturers and becomes an industry standard, so that play options are now Seated, Standing, or WRIP (walk/run in place). Having spent enough time with CAOTS to realise how effective it is, I would always choose WRIP wherever larger-than-boundary traversals are desired… which let’s face it, is almost all the time!
Alternatively if FLVR could be packaged as an add-on or patch controller configuration, to work with any controller or game, that would be dandy. However my attempts to use VRocker suggest that trying to be compatible with every game is a hard target. Having game devs adopt FLVR at the SDK level to incorporate into their devel process seems like a better bet.
The Room VR, to name just one, would be twice the game it is if it were freely navigable by walking, instead of limited to teleportation to a small set of fixed locations. Vader Immortal, H3VR, Google Earth VR, Sports Bar VR, Nature Treks VR, Walkabout Mini Golf… the list of games/apps that would benefit from “real walking” to move around is very long. Point-and-snap teleportation is just an ugly, immersion-destroying hack. A more natural method of locomotion could help popularise VR with the general public by reducing “VR seasickness” and making it easier and more intuitive to enter simulated environments (no gamepad-like controls to memorise).
This is very fine work; the developer’s skill and attention to detail are noticeable throughout. If you are trying out the demo, be patient; I didn’t really get it tweaked until I was past the model town and into the “rooftops;” and I didn’t realise the true joy of the locomotion method until I got to the “logging road” demo at the very end.
That logging road sim was what wowed me; earlier phases of the demo were impressive, but it wasn’t until I got out into a biggish open world with great scenery, long sightlines, and a generous amount of road/trail ahead that I really felt the full awesomeness of this locomotion method. I jogged for many minutes around the frozen lake and was more immersed than I’ve ever felt in a VR environment. Not having to (unnaturally) use buttons and sticks to move around is more powerful than you can imagine if you have not yet experienced it! This is how I want to move through VR open worlds.
Huge props to the dev (George Kong) who is imho a genius and a trailblazer. I wish I knew how to get the industry to adopt this code. It would push VR one big step further towards market saturation for sure! [end of review]
Development continues, as VR enthusiasts and the industry try to figure out how to “get our legs under us” again. Who knows which solution will prevail, but I’m really hoping for something as much as possible like Freedom Locomotion. The immersive euphoria of running at a realistic speed through a realistic VR landscape is hard to convey in words, but it made a profound impression on me; it leaves me with a lingering sadness every time I have to glide or teleport to get around in an otherwise deeply engaging VR game or experience.
I invite comments and discussion; if you’ve been very satisfied with solutions that I dismissed, please share your experience. If you know of other nifty ways to walk or run in VR, please let me know about them.