Ever wonder how VR tracking systems really stack up? A research and engineering group called the Virtual Dimension Center (VDC) in Fellbach, Germany conducted a study recently that reveals just how accurate some of the top VR headset’s are when it comes to tracking their respective motion controllers.

When most people compare VR tracking accuracy, you typically get a response about how it feels in relation to a known quantity, like when you play Beat Saber across two different headsets to see how they match up. That works when trying to help someone understand how they might perceive the overall experience, but it does little in the way of objectively proving the point.

According to the test conducted by VDC Fellbach, HTC Vive Cosmos’ inside-out optical system consistently shakes out as the least accurate of the bunch.

Note: This article refers to the stock HTC Vive Cosmos, and not Vive Cosmos Elite, which ships with a SteamVR tracking-compatible faceplate.

How It Was Measured

As first reported by German publication MIXED (German), VDC Fellbach measured the tracking systems on Oculus Rift S, Oculus Quest, Windows Mixed Reality, Valve’s SteamVR (aka Lighthouse) tracking, and HTC Vive Cosmos. Ostensibly as a sort of control, the study also included the industrial-grade tracking cameras from ART, which routinely outperformed even the rock-solid external SteamVR laser-based tracking system.

The objective was to quantify each system’s input accuracy, measured in millimeters, which was done by attaching each controller to a robot arm with an adaptor. A 3D-printed head model was used to hold the target headset while the robot arm moved the VR controllers 50 times between two points on the X-axis, set 500 millimeters (~20 inches) apart, or about an arm’s distance away.

Image courtesy VDC Fellbach, translated by Road to VR

The robotic arm, which was connected to the laptop via LAN, was controlled by an in-house software using a modified variant of OpenVR. This allowed the testers to not only reliably know the actual position of the controller, but also the reported position of the controller, capturing data in the X, Y and Z-axis during each headset’s trial.

SEE ALSO
Analysis: Steam Hangs on to VR Surge Four Months After 'Half-Life: Alyx Launch'

The Results

Below you can see the standard deviation of the results for position one, which is closest to the headset. The study also includes the maximum deviation and mean, however they all point to more or less the same relative conclusion.

Image courtesy VDC Fellbach, translated by Road to VR

SteamVR tracking, which uses laser-based ‘outside-in’ beacons, is the most accurate consumer-grade solution of the bunch. Notably, ART’s outside-in optical tracking is so accurate that its bar isn’t even visible on the far right of each chart.

Windows Mixed Reality performs better than we expected in position one, coming in neck-to-neck with Oculus Rift S. HTC Vive Cosmos lags behind by a wide margin with its 2-3 mm deviation. This changes somewhat when the respective controllers move to position 2 however.

Image courtesy VDC Fellbach, translated by Road to VR

Outside-in tracking remains more or less the same regardless of the distance between the headset and the controller, however inside-out tracking takes a notable hit in accuracy as it moves farther away from the headset.

In one instance, Windows Mixed Reality underperforms by nearly 3 mm from dead zero, beating out Vive Cosmos by nearly a millimeter in accuracy.

Robot arm and controller adaptor | Image courtesy VDC Fellbach

The charts above don’t tell the whole story though. According to the study, HTC Vive Cosmos performs “worst,” with deviations that are “sometimes above 5 mm,” and that it was even possible to visually detect Vive Cosmos’ deviation while the measurement was being carried out.

Granted, Vive Cosmos has benefitted from multiple software updates since its late 2019 launch, which was so bad at the time that we actually postponed our initial review until HTC addressed our concerns over tracking. According to this study though, it seems that well into late 2020 that Cosmos is still underperforming in the tracking department.

The testers admit their findings may be influenced by “differences in the hardware components used and different software,” however it’s unlikely a research group capable of creating such a testing platform in the first place would simply forget to update their drivers. At least we’d hope not.

– – — – –

If you speak German, or don’t mind flexing your Google Translate skills, you can check out the full paper to see VDC Fellbach’s findings for yourself. A special thanks goes out to our friends at MIXED for pointing us to the news.

Newsletter graphic

This article may contain affiliate links. If you click an affiliate link and buy a product we may receive a small commission which helps support the publication. More information.


Scott Hayden
Well before the first modern XR products hit the market, Scott recognized the potential of the technology and set out to understand and document its growth. He has been professionally reporting on the space for nearly a decade as Editor at Road to VR, authoring more than 4,000 articles on the topic. Scott brings that seasoned insight to his reporting from major industry events across the globe.
  • y_m_o

    Inside-Out tracking has a long way to go. Think I’ll stick with my Lighthouses tyvm.

    • Dan Lokemoen

      Yeah, tell yourself that, during a game of BeatSaber, you can tell the difference between deviations of .1mm and .3mm.

      • silvaring

        The deviations on the Z-axis at a distance were quite a lot more than .1mm though. For example stretch your arm out and get within 5mm of an object with your finger… basically your body should be able to hold that position without touching the object. Unfortunately in WMR tracking this error is the difference between contact and no contact. The margins are that tight! Anyway I agree with you wholeheartedly for the most part, the tracking in WMR is certainly fantastic all round, and good enough for most use case scenarios.

        • R3ST4RT

          TBH, the reason I play beatsaber on my Index is because of the tracking volume. I find that when I get really into things, my hands leave the tracking volume of the oculus quest quite frequently. It’s much less about tracking quality and more about volume and consistency.

          • silvaring

            I think there’s definitely something to be said for tracking ‘feeling’ better without being able to quantify exactly why or how. Not everyone has the same sensitivity to this, and its definitely a case of some people being more sensitive to it than others. E.G Jet Fighter pilots have a very acute sense of FPS… much more than the average person.

      • JDawg

        I own the Index and the Quest. There is a huge huge different between the two during Beat Saber and other games. Index is so much more accurate when moving fast and slow. Especially when moving fast because the Quest will lose tracking frequently for short moments.

        • Dan Lokemoen

          The Quest is a huge downgrade from PC-based VR, so no surprises there. I have only ever used my WMR HMD, so nothing to compare, but I always suspected complaints about the tracking were out of touch. I wonder if WMR software was a bit crap when it first came out and they fixed a lot of things in the first year.

      • I installed my base stations on my wall mounts from my CV1 cameras, over a year ago. Not touched them since. I just turn on the mains power socket below each mount, no pain in butt at all, just class leading tracking with no occlusion issues

        • Dan Lokemoen

          Depends on the person. For many people, setting up the lighthouses is just something they are never going to do. Also, lighthouses apparently glitch once in awhile and lose tracking of the player’s head (maybe this is fixed?) — this NEVER happens with WMR.

      • TechPassion

        TRUE. Screw with sticking of boxes on walls and polluting rooms with cables.

  • Dan Lokemoen

    Many people deride WMR’s tracking, while I have always thought it was really good. Even considering one poor mark, it looks like WMR punches in with Oculus. When reading this chart, realize that almost all of these deviations are less than a millimeter (and the ARTT scores so well that its color doesn’t show up on the chart).

    • asdf

      um no lol

      • Dan Lokemoen

        um yeah lol

      • johann jensson

        He said basically, which means that in his use case there would be no difference.

    • Stefan

      Note that the study only measures accuracy in an optimal position for forward facing cameras. The difference between Lighthouse and WMR lies mainly in tracking volume. Also, tracking quality will probably be alot worse closer to the tracking volume limits.

    • Kevin White

      Over time, I noticed more latency in WMR (original model Odyssey). For example, in Longbow (The Lab), where I generally didn’t have tracking volume or proximity to headset issues, I nonetheless had a more disconnected feeling. I “kind of” got used to it over about 10 hours of trying / testing. But then I went back to a solid Lighthouse setup, and the difference was pronounced — I felt like I didn’t have to compensate with the Vive wands, things felt more immediate and true, and my performance was immediately much better in the game.

      Another application that makes it clear is the “whack-a-mole” minigame in the nVidia Funhouse from a few years ago. Try it. The “mallet” feels more solid and connected with Lighthouse than with WMR.

      • Dan Lokemoen

        I have never seen a test of latency and I don’t own a pile of HMD’s, so it’s pretty hard to say. I certainly don’t think WMR is perfect or the best — it’s just WAY better than most people seem to think and has some unsung advantages like more joysticks on the controller and emulated roomscale.

  • xyzs

    Inside-out can only improve and get closer and closer to lighthouse precision. Very soon that will be completely pointless to install such a cumbersome system.

    • Rogue Transfer

      It’s not the precision, but the tracking volume and less likely to have the the controller(or arms) block the view of the other controller.

      Because the Valve SteamVR tracking has ~20 sensors inside each controller looking out in all directions for the IR light from the Lighthouses, they can work even behind the back, where cameras-on-headset tracking can’t ever see through to.

      Cameras-on-headset tracking would require fusion with another tracking technology, like electro-magnetic or ultra-sonic tracking to fill in for when tracked devices go out of view for longer or get blocked. Less accurate than SteamVR tracking, but would be sufficient then to solve the dead spots(and then even better for occlusion proofing, at a loss of accuracy outside view).

        • Schadows

          To be honest, what is shown in the picture isn’t specific to the tracking with base station.
          It could be done with the other inside-out tracked headset with even much more freedom (base-station 2.0 are limited to 4 unit if the description on Valve site is correct).

          • Where WMR and Oculus insight don’t fare so well is in featureless spaces or with variable light levels (big windows with reflected sunshine from River Thames in my image)

            1 setup of steamVR allows multiple users in empty space with variable light levels and tracked accessories, previously the realm of Optitrack. It won’t scale like Optitrack but it’s a fraction of cost and rigging only takes a few minutes.

            I found my WMR system worked best with window blinds shut and mains light on to maintain consistent light levels.

          • Zantetsu

            But if you were willing to put the effort into setting up and configuring lighthouse base stations, you could have been willing to put the effort into putting pieces of tape on the walls or whatever it is that is needed to improve Insight’s ability to detect features needed for tracking.

            I am a lighthouse fan but I got an Oculus Quest recently and I feel like it’s tracking is ‘good enough’ for most scenarios. If I had a big room with flat white walls I would just put some decals on.

          • Agree the Quest tracking is good enough, as was WMR headset in well lit room (controllers not so impressive). But steamVR is robust, accurate and easy to setup, lighthouses are awesome.

            “Lighthouse Towers” from ‘Roomscale Plus’ experiments in early 2017, Vive running off PC on wheeled mechanics trolley and long power cable drums. Allowed setup in number of empty rooms of different size without having to modify the room interior or do anything with lighting.

            https://uploads.disquscdn.com/images/d106b4c795d60ccba3523a46dcb8f1689aa0f5d83791c88023ec088b0440afce.jpg

        • Dan Lokemoen

          Most HMD’s do SteamVR. I don’t think you’re saying what you mean to say.

    • Ad

      Lighthouse tracking allows;

      Accessories like body trackers and VR pens which could be crucial to using VR for work

      Modularity. Buy them once and use them with all your headsets, and mix and match controllers and accessories with headsets. Makes it feel like a better investment. You can even technically use steamVR tracked objects without the headset if you want.

      The most accuracy independent of your body. I can do anything without worrying in the slightest where my face is looking or how close the controllers are to my face.

      Also worth noting that SteamVR 2.0 has such a wide FOV that you don’t actually need to mount them on your wall. And Valve could make the process really easy by having you make your boundaries inside the headset with pass through, I don’t know why they haven’t yet.

    • Schadows

      I would say the inside-out “camera-based” tracking of the headsets is already pretty good.
      What is lacking is the outside-in tracking of their controllers (the headsets are tracking the controller, the controller aren’t tracking their own position like the valve ones).

      Sure they will improve that aspect with time, but I’m still skeptical about the shortcoming of the tech, especially occlusion from the body, of the other controller.

      Wet and sea !

  • Schadows

    Just to be clear, Valve Index (and HTC Vive) are Inside-out headset too.
    Basestation are not camera and they are not tracking where the headset is (unlike OG Rift). It is the headset which is tracking where are the basestation (or more precisely where are the infrared beam emited by the basestations).

    • Kevin White

      Yah, if we’re just talking about controllers, then Lighthouse controllers (Vive Wands and Index Controllers) are really the only inside-out tracked controllers from the perspective of the controllers themselves. They just happen to need laser beacons.

      Quest controllers, Rift S controllers, Cosmos controllers, WMR controllers — all tracked outside-in from the perspective of the controllers, by cameras that are external to the controllers. Rift CV1 Touch and PSVR Move are tracked outside-in by stationary cameras.

      What I would like to see, where I think we need to go (though I don’t know if it is technically feasible, is in true inside-out tracking for controllers, without laser beacons. Basically, bring the controllers up to parity with the inside-out tracked head-mounted units by mounting the cameras ON the controllers themselves, so they track themselves, using features of the walls, floors, and ceiling rather than laser beacons like Lighthouse controlers must.

      • Schadows

        Replacing laser-sensor on the controllers by cameras to track the environment by themselves (which should eliminate any oclusion issues we can encounter with Quest/Rift S and the like) would be nice indeed. But I’m not sure it is feasible right now because of the size and number of cameras capable of tracking the environment accurately that would be needed.

        I have (personally) no problem with base-stations, but I can understand how easier it is to just play wherever you want (almost) without them.

        • Ad

          I really hope Valve can get the price for lighthouses down because I worry that camera tracking will take over making it cheaper but less able to support accessories and modularity.

        • Mradr

          “Number of cameras” you mean like 2? You only need 2 per controller if you go inside out tracking from the controllers. You then use the headset as the anchor just like before when insight.

          • Schadows

            2 wouldn’t be enough to get reliable and accurate 3d/spatial awareness without the risk of occlusion.

    • DeenVR

      Pedantic. Nobody cares about these semantics – camera based tracking has become synonymous with inside-out and everyone knows what that means.

      • Schadows

        It matters in this case because the article is about comparing headset technologies, and making the conclusion that outside-in is more precise is just false (all tested headsets are inside-out), especially since, like @disqus_c1gLmiDKD9:disqus explained, from the controller point of view, this is exactly the opposite (only Valve engineered controllers use inside-out tracking).

        If the conclusion was “camera based tracking” is better/worse then, indeed, there would be no reason to point that out.

        PS : and just because it has become synonymous, doesn’t mean it’s right to keep propaging this false fact.

        • dogtato

          it matter but you’re mistaken. It’s not about where the camera sensor is, it’s about performance characteristics. The lighthouse system operates like an outside-in system. The further you get from the lighthouses the less accurate. For inside-out systems, they are less accurate the further you get from the headset. This article shows that well. That is why lighthouses are outside-in despite the sensors being on the headset.

          • Nickel Dare

            So if my car performs like a ferrari, it is a ferrari?
            That is quit the flawed logic. Lighthouse is Inside-Out tracking and there is nothing to debate about it. Inside-Out means the Logic happens in the device, which is the HMD and this is accurate. The base stations are dumb, they do not know where you are nor do they care, all they do is emit a IR laser which your device picks up and determines its position in 3D space. It is not talking back to the base station in any way thus it is Inside-Out.

          • dogtato

            That’s logical enough, but it doesn’t matter which part is the sensor. Researchers classify lighthouse as outside-in because the way it works is fundamentally the same as how it works when you have an external camera looking at a headset with lights on it. The lighthouses shoot out a spread of rays that hit the sensors on the headset, and precision declines as you move away from the lighthouse. A camera receives a spread of rays from some point sensors. In both scenarios you are locating the headset by determining where it is in the field of view of the external camera/lighthouse.

          • Schadows

            You’re right about loosing the precision the further away you get from the base-station, at least in theory (since this isn’t tested here), but that still doesn’t make lighthouse an outside-in tracking (which is only based upon who’s tracking who).
            But indeed, you see that real outside-in tracking (the controllers for all the others headset which are tracked by the headset) show more deviation the further the controllers get from the camera tracking them (on the headsets) … just like you said.

            It would have been interesting to see at which distance from the base-station the accuracy become equal or worse than outside-in tracking of the other controllers (knowing that the maximum tracking distance for non-lighthouse based controller is the length or our arms anyway).

          • Nickel Dare

            https://www.researchgate.net/figure/The-inside-out-tracking-system_fig2_2640852
            There you have a research paper detailing “Tracking a Head-Mounted Display in a Room-Sized Environment With Head-Mounted Cameras”. I cite a part of the paper:
            “This paper presents our efforts to accurately track a Head-Mounted Display (HMD) in a large environment. […]Both systems use an inside-out optical tracking scheme […]”.

            Now please show me your “Researchers”. And I want actual papers, texts etc. not some Wiki Pages or forum quotes. Go ahead, I will wait for that even though I know you won’t find any.

            I stand by my point. Lighthouse is Inside-Out, Researchers and Developers alike call it that because it works exactly like a Inside-Out technology works. Inside-Out and Outside-In is NOT about when something loses accuracy or not, it is about where the computational logic is that calculates your position in 3D-space. Outside-In means your Headset is dumb and the surrounding determines your position and Inside-Out means your surrounding is dumb and the HMD determines its position. The fact that lighthouses only shoot laser beams and do no computation at all means this system is a Inside-Out system. I am sorry but that is the way those technologies are classified no matter how much you like or dislike it.

          • Nickel Dare

            Update for you.

            I have contacted the author of this research paper that was discussed here and asked for clarification. He said he will update the paper to reflect the correct choice of words as it is, as he said it himself, indeed “Inside-Out”.

          • Kevin White

            Excellent.

          • dogtato

            In the early days of steamvr I argued that it was obviously inside-out because of all the obvious reasons, but I was linked to research paper and a very careful reading of it said that lighthouse was outside-in. The reasoning was the way it worked, by tracking the angles between base-stations and the headset, was just a trivial swap between the sensor and the emitter and didn’t really change how the system worked except it was clever.

            But at this point, the logic of “the sensor is on the headset looking out” has confused enough people that “inside-out” is starting to win out even though the way it functions is much different than how every other “inside-out” system functions with computer vision.

            What really matters is the algorithm, not where the sensor is. The advantage of lighthouse over an external camera looking at lights is precision and tracking volume, but the math of how it tracks is the same as other outside-in tracking.

          • Nickel Dare

            See that’s where you are wrong by assuming that because it is not like “the other Inside-Out”, it must be “Outside-In”. The thing is, there is not just 1 Inside-Out tech out there. WMR and most other Inside-Out headsets are so called “Markerless” Inside-Out, where the HMD has cameras and tries to pinpoint its position by tracking geometry around it. Lighthouses act like “Markers” in that they are fixed positions the HMD can orient itself and depending on where they are in relation to it, the HMD calculates it’s position more accurately. That’s why it’s called “Marker based Inside-Out” tracking. It is still Inside-Out tracking, because the HMD (Inside) is looking and orienting itself on the World around it (Out) thus it is Inside-Out.

    • psuedonymous

      No, Lighthouse is an Outside In system. Whether a system is inside-out or outside-in has nothing whatsoever to do with which direction photons are flying in the system. It’s to do with how the coordinate system works. “Where the cameras are” is an oversimplified rule-of-thumb that does not apply universally and confuses laymen (especially in systems with no cameras at all)).

      An Outside-In system has the coordinate system of the tracking system fixed relative to the world coordinates.

      An Inside-out system has the coordinate system of the tracking system move relative to the world coordinates, and fixed relative to the tracked object.

      Lighthouse, like the ArcSecond system that predates it, is Outside-in because all its coordinates are fixed relative to the world: each tracker only knows it’s pose as the relative angle (angular position in polar coordinates) relative to each laser beam sweep of a basestation. i.e. it has an X and Y cooridnate relative to the basestation. This is exactly how a camera system also works: you get an X and Y coordinate of a marker (illuminated or passive) relative to the camera origin (the camera system will also be using polar coordinates once you characterise the camera’s intrinsic to perform resectioning, as is necessary for accurate tracking).

      In summary: tracking of controllers from an Outisde-in system (Lighthouse, CV1 Constellation) are both trivially Outside-in problems. Tracking of controllers for Rift S, Quest, and WinMR, and Cosmos are two stacked systems due to the double coordinate transform: first an Inside-out system to get the HMD pose in world coordinates, then an Outside-in system to get HMD-relative coordinates for the controllers. Because you care only about world-coordinates for your controllers (both for rendering, and for sensor-fusion with the world-relative IMUs), that means these systems will necessarily compound all positioning errors due to the double transform.

      This was all common knowledge before Lighthouse even existed. Check the P.30 inset box here ( https://web.archive.org/web/20170625030334/http://www1.cs.columbia.edu/~drexel/CandExam/Motion_Welch_Foxlin.pdf ) describing the arcSecond system, now called Nikon iGPS, which is a scanning laser system that is identical to Lighthouse in operation but predates it, and is used for precision metrology. That article is from 2002.

      • Nickel Dare

        Lighthouse is Inside-Out.
        Inside-Out and Outside-In define where the calculating logic sits.
        The Lighthouse stations are just IR light emitters. They have no logic, they are just there to send out a IR laser in fixed timesteps.

        The HMD pics up those lasers and because it knows that those things don’t move, it can calculate it’s position relative to where the IR light is coming from. People mix this up all the time. In a “outside-in” system, the HMD would be sending out IR or other kinds of light and the Stations in the room would pick this light up and calculate the HMDs position.

        Lighthouse is a marker based inside-out system.
        WMR is a markerless inside-out system.

        If you’ve have read the thread further, you would have seen that I have already contacted one of the authors of the paper that is cited here and he confirmed to me that Lighthouse is indeed Inside-Out and he is going to reflect this in a update to the paper soon.

      • Schadows

        “An Outside-In system has the coordinate system of the tracking system fixed relative to the world coordinates.

        An Inside-out system has the coordinate system of the tracking system move relative to the world coordinates, and fixed relative to the tracked object.”

        And that’s exactly why lighthouse is an inside-out tracking system.
        Lighthouse doesn’t track anything (doesn’t have that capabilities).
        Lighthouse doesn’t have any coordinates awareness nor any logic implemented on their side.
        Lighthouse are just IR emitters which will then serve as beacons for the headsets and the controllers to compute their position, the same way the non-lighthouse based headset used external shapes and contrast to do theirs.
        The headset and controllers are tracking the IR rays in the environment emitted by the lighthouse, not the other way around.
        The coordinates of the tracking system has its origin at the headset (and controllers), hence moving relative to the world.

        As for the article you quoted, you should note that nowhere do they qualify the VMS from Arc Second as an pure outside-in system. They even acknowledge that the system works as the “inside-out” definition is, but that it is “misleading” because one point in the tracking system, the orientation sensitivity, is more commonly seen in outside-in tracking.
        Sure, we can argue that the tracking is also made possible with the timing of the IR swipes, which is an external factor, but in the end, the lighthouse has no knowledge of the tracked target. It just acts independently and constantly the same way, making it a part of the statically tracked environment.

      • Kevin White

        “…confuses laymen…”

        See, the problem is we are enthusiasts, not laymen, and we ought not dumb things down for laymen. It’s like an automotive enthusiast forum just calling a CVT “automatic” because they don’t want to confuse lay people, even though a CVT is distinctly different from a traditional automatic with viscous coupling and a torque converter.

        We should be free to use exact language here, especially on a technical article.

        The Vive and Index HMDs, and the Vive wands and Index controllers all track “themselves” using stationary external laser beacons. The WMR, Quest, Rift S, and Cosmos HMDs all track “themselves” using stationary features of your walls, ceiling, and floors. Everything in this paragraph uses inside-out tracking.

        The WMR, Quest, Rift S, and Cosmos controllers are all tracked by cameras that are external to them. The PSVR HMD and the Rift CV1 HMD are tracked by cameras that are external to them, and the PSVR Move controllers and Rift CV1 Touch controllers are also tracked by cameras that are external to them. Everything in this paragraph uses outside-in tracking.

        I get it. We want to say WMR, Rift S are inside-out systems because you don’t have to set anything up, provided your room isn’t featureless. They are actually hybrid systems, with their HMDs tracking themselves, but with their controllers being tracked by the HMD’s cameras. But in a technical article, we should say the controllers are not inside-out tracked — that is actually something we haven’t seen yet, controllers that can track THEMSELVES using features of the walls, ceiling, and floor. The outside-in nature of the tracking of the controllers for these systems is where their limitations stem from. I’m an advocate for inside-out for all three (HMD, controller left, controller right), which Lighthouse has, and I’d love to see markerless inside-out tracking brought to controllers (or maybe some hybrid system where the HMD’s cameras track the controllers in conjunction with controller-mounted cameras).

        Most of the time it seems these terms are being used while conflating actual tracking schema with concepts such as “portability,” “ease of setup,” and “amount of damage you have to do to your walls.” No argument from me that Lighthouse (and Rift CV1) is less portable, more taxing to initially set up, and can require drilling into walls. But we’re not done yet — we need the controllers themselves to also have inside-out tracking before we have truly excellent tracking for all three essential tracked components (HMD, 2 controllers) while having easy portability and setup.

      • Kevin White

        “…confuses laymen…”

        See, the problem is we are enthusiasts, not laymen, and we ought not dumb things down for laymen. It’s like an automotive enthusiast forum just calling a CVT “automatic” because they don’t want to confuse lay people, even though a CVT is distinctly different from a traditional automatic with distinct gears, viscous coupling, and a torque converter.

        We should be free to use exact language here, especially on a technical article.

        The Vive and Index HMDs, and the Vive wands and Index controllers all track “themselves” using stationary external laser beacons. The WMR, Quest, Rift S, and Cosmos HMDs all track “themselves” using stationary features of your walls, ceiling, and floors. Everything in this paragraph uses inside-out tracking.

        The WMR, Quest, Rift S, and Cosmos controllers are all tracked by cameras that are external to them. The PSVR HMD and the Rift CV1 HMD are tracked by cameras that are external to them, and the PSVR Move controllers and Rift CV1 Touch controllers are also tracked by cameras that are external to them. Everything in this paragraph uses outside-in tracking.

        I get it. We want to say WMR, Rift S are inside-out systems because you don’t have to set anything up, provided your room isn’t featureless. They are actually hybrid systems, with their HMDs tracking themselves, but with their controllers being tracked by the HMD’s cameras. But in a technical article, we should say the controllers are not inside-out tracked — that is actually something we haven’t seen yet, controllers that can track THEMSELVES using features of the walls, ceiling, and floor. The outside-in nature of the tracking of the controllers for these systems
        is where their limitations stem from. I’m an advocate for inside-out for
        all three (HMD, controller left, controller right), which Lighthouse
        has, and I’d love to see markerless inside-out tracking brought to
        controllers (or maybe some hybrid system where the HMD’s cameras track the controllers in conjunction with controller-mounted cameras).

        Most of the time it seems these terms are being used while conflating actual tracking schema with concepts such as “portability,” “ease of setup,” and “amount of damage you have to do to your walls.” No argument from me that Lighthouse (and Rift CV1) is less portable, more taxing to initially set up, and can require drilling into walls. But we’re not
        done yet — we need the controllers themselves to also have inside-out
        tracking before we have truly excellent tracking for all three essential
        tracked components (HMD, 2 controllers) while having easy portability
        and setup.

  • doug

    Great VR tech journalism.

  • doug

    It’d be cool if the Germans (or roadtovr) used a rig to map tracking volume.

    • steamVR tracks a surprisingly large volume. I got it working with 9 metres diagonal gap between 1.0 base stations using sync cables, and have used 4 base station set-up in huge space at VR festival with backpack PC.

      Optitrack for the warehouse scale setups though was an amazing experience again with backpack PC’s.

      • Ad

        I still haven’t seen a clear guide to how well laptop VR actually works. I’ve seen one LTT video where he puts a wireframe backpack on and sticks on a laptop (and uses a vive instead of WMR for some reason) but that’s it.

  • Rogue Transfer

    So, since you say the IR laser emitters are ‘outside-in’ beacons, does this mean you think because the WMR & Rift S that require ‘outside-in’ light from light bulbs in the room to track by, that they too are ‘outside-in’ tracking? They both need ‘outside-in’ light to do headset tracking.

    Outside-in and inside-out are specific, technical tracking terms to define where the sensors are and the direction to the tracked device – not the opposite. The Lighthouses don’t have any tracking sensors and don’t track devices, so they don’t fit the technical tracking term of outside-in.

    It’d be better to avoid twisting the technical terms, but use simpler, lay person terms like external beacons(Lighthouses) or cameras-on-headset(WMR, Rift S & Cosmos). After all, anyone new to VR won’t have a clue what you mean and will get mixed information and corrections from other sources.

    • Ad

      I think it’s because the ART system actually is outside in. But yes, people don’t understand why they’re called “lighthouses” clearly.

    • psuedonymous

      “Outside-in and inside-out are specific, technical tracking terms to
      define where the sensors are and the direction to the tracked device –
      not the opposite”

      Nope, a common misconception, see my post above. Outside-in and Inside-out refers to how the coordinate transform of the tracking system works, and is independent of the sensors used, cameras or otherwise. You can have an Inside-out or an Outside-in magnetic tracking system with no cameras whatsoever, the distinction is whether the tracking system coordinates are world-fixed or tracked-object-fixed. Lighthouse is Outside-in because the coordinates are all world-relative (measured beam sweep angle relative to the fixed base stations).

  • With all respect for the researchers, but I got to the same conclusion without using a robotic arm. I have a Cosmos, and while the positional tracking of the headset has improved a lot in the various updates and now it is good, the controllers are still bad. You can watch on Youtube my play session with HalfLife Alyx on the Cosmos. I couldn’t shoot anything with precision

  • Ad

    Valve should have leveraged this a bit more for things like mixed reality or something, or releasing their own version of the steamVR pen for consumers. But I’m surprised the quest is beating the rift S on this.

    Cosmos uses an RBG camera, that’s why it’s so bad, right?

  • DickDastardly

    If Google Translate is to be believed, the authors don’t seem to mention where the lighthouses were positioned, making replication of the study impossible. There is a line where they say:

    “Das zu testende HMD bzw. das HMD zum jeweiligen Trackingsystem wird auf eine eigens an-gefertigte Halterung an der Tischkante mit direktem und freien Sichtfeld zu den Motion Con-trollern angebracht.”

    ..which Google Translate gives as:

    “The HMD to be tested or the HMD for the respective tracking system is attached to a specially made bracket on the edge of the table with a direct and free field of view to the motion controllers.”

    If the second “HMD” mentioned actually refers to a lighthouse then the test is a very artificial one, conducted far closer to the lighthouse than a typical user with a ceiling or wall mounted setup would ever get (unless stood in the corner directly under the lighthouse and reaching up towards it).

    They also placed the heavy moving robot arm and the HMDs (strapped to a model head) on the same table which seems like an odd choice given the possibility that the movements of the robot might cause the HMDs to settle slightly on the head, introducing significant errors in the measurements on the sub-mm scale in question for the non-lighthouse (or ART) HMDs.

    Meanwhile, if lighthouses weren’t on the table then they’d have been totally unaffected by movements of the robot. On the other hand, even if they were connected to the table via “a specially made bracket” (presumably using the standard tripod screw connector) then they’d have been far less likely to move from their initial position after the table is wobbled several dozen times than HMDs strapped to a model head sitting on (rather than attached to) the table. In other words, all the study might show is that brackets and screws provide a more rigid connection than foam and straps.

  • Ardra Diva

    This is where Oculus is clearly superior. hand controllers.