Sixense, the company behind the much-belated positionally tracked STEM controller system, just released an update stating the company is “getting close to be able to start production,” which should follow an estimated 4-week retooling and finalization of production samples.
As the result of a successful Kickstarter campaign launched back in late 2013, STEM was poised to beat many manufacturers to the punch by offering the first consumer room-scale controller and positional tracking systems using a magnetic tracking technology. After more than 4 years of waiting, Sixense CEO Amir Rubin says that next week the company should receive updated 3D printed samples following a recent design update.
“Assuming that the fix works, the tool change and samples, should not take more than four weeks,” Rubin says. “We are getting close to be able to start production, so please stay with us through this final phase.”
The company’s penultimate update detailed the outstanding design issue – a critical misalignment and part warping problem that prevented STEM from entering production phase. Taking the words out of many backer’s mouths, Sixense called the effort “a long and complicated process.”
Here’s the full update below:
Update – March 2018 Posted by Sixense (Creator)
We have received an update from our manufacturer in regards to the remaining issue concerning the STEM plastics.
The fix that is proposed, and that we will be implementing, incorporates two modifications. The first mod, shown in the images below, is the addition of a screw boss near the thumb area where we were experiencing some inconsistency in the reveal gap.
image courtesy Sixense
image courtesy Sixense
The second modification includes additional alignment/interlock features around the perimeter of the two halves that will help stabilize the reveal all the way around the assembly.
image courtesy Sixense
image courtesy Sixense
We should be receiving 3D printed samples of the updated geometry next week, before we initiate the tool change to do an assembly level check fit to ensure the design update. Assuming that the fix works, the tool change and samples, should not take more than four weeks. We are getting close to be able to start production, so please stay with us through this final phase.
Also, we are exhibiting our latest VR products at GDC again this year. If you plan on attending, we welcome you to come visit us in Booth #1201 in the South Hall of Moscone Center.
Thanks, Amir
Sixense originally posted their first delay back in April 2014, citing the need for unspecified “hardware improvements.” Similar delays thereafter lead up to one of the greatest setbacks to the project in late March 2015: failure to pass FCC/CE testing, something that was chalked up to difficulties surrounding the basestation’s inability to both provide grounding for its radio frequency (RF) dongles without disturbing the magnetic tracking. Traversing more failed tests and manufacturing difficulties, a July 2017 update revealed a manufacturers letter to Sixense that detailed the company was still having part stability issues.
While many might wonder exactly why a company would sally forth to produce a product largely thought surpassed by modern VR systems such as Valve’s SteamVR tracking standard or Oculus’ optical tracking sensors, STEM at very least boasts a function that other positional tracking systems can’t claim: no need for direct line-of-sight from the basestation. Whether that’s enough of a draw, we can’t say for sure.
So here we are again – on the hypothetical cusp of Sixense STEM finally materializing into an honest-to-goodness product.
Get Schwifty!
This business of no LOS is actually very big deal, as both Rift and Vive have tracking issues even under the best of conditions due to loss of LOS. It may well be that Rift uses the least effective (but still good) tracking method, Lighthouse better, this approach might well be the best alternative with refinement. Curious now to see where this goes, I was not clear on the magnetic tracking before.
Caven
Even if it turned out not to be a compelling replacement for Lighthouse or Constellation, it might ve very useful for mobile VR, where occlusion is a harder problem to deal with.
Sponge Bob
Magic Leap AR glasses uses magnetic tracking for hand held controller
It remains to be seen how good they are
better be good after spending so much money
mirak
No it’s not usefull for mobile vr because you need a base station to create the magnetic field.
Caven
Not a problem. Simply build the base station into the headset. That’s what the Windows Mixed Reality headsets do with their cameras to avoid the need for stationary cameras like the ones used by the Oculus Rift.
mirak
No it won’t work, because the Stem base station is an emitter, like the Vive lighthouse, except one emit magnetic field, and the other emit light.
They are not receivers, unlike cameras.
They are not radars either.
Caven
I don’t see where that’s a problem. In the case of Windows Mixed Reality headsets, neither the emitters (the controllers) nor the receiver (the HMD) are stationary. The controllers only need to be tracked relative to the headset, so a headset incorporating a Stem-style system can use cameras to track the HMD in world space, and the Stem system to track the controllers relative to the HMD. Combined with IR emitters on the controllers for primary tracking, the system could conceivably keep calibrating the magnetic field in real-time, to ensure maximum accuracy when the magnetic field is needed to overcome occlusion issues.
mirak
You initially didn’t say you were thinking about it only for controlers tracking.
Sponge Bob
No, its not a replacement – can’t be because of physics laws
magnetic field attenuation is inverse 6th power of distance
thus tracking distance is limited to few feet at best – not even close to e.g. 10 feet
that’s all you need to know
Get Schwifty!
What range do they specify? I was not unaware of the range questions being college educated in technology, but this also somewhat depends on sensitivity, etc. Considering certain animals utilize magnetic sensitivity to navigate, etc. over fairly large distances with fine detection, its not out of the bounds of possibility to use it for tracking 30+ feet with the right technology.
Sponge Bob
you do understand how weak magnetic field is at 10 feet ?
any small electric current going through some wire nearby will generate its own magnetic field that will completely mask the target
room scale is possible with big coils and large power consumption
industrial applications – yes, may be
home gaming setup – not practical at all
Sponge Bob
magnetic
optical
acoustic
any other way to skin the cat you know about ?
M Rob
Wow they added a screw!!!…..I’m so super pumped for this project.
Ballpeen
Lol, surprised to hear they are still having design issues 4 yrs after promised delivery. =:-o
Sponge Bob
those are not design issues
those are basic laws of physics
can’t change those
Ballpeen
I think you may be referring to the issues they had in 2014. I was referring to the changes:
“The fix that is proposed, and that we will be implementing, incorporates two modifications. The first mod, shown in the images below, is the addition of a screw boss near the thumb area where we were experiencing some inconsistency in the reveal gap.
The second modification includes additional alignment/interlock features around the perimeter of the two halves that will help stabilize the reveal all the way around the assembly.
We should be receiving 3D printed samples of the updated geometry next week, before we initiate the tool change to do an assembly level check fit to ensure the design update.<===="
Sponge Bob
do you want them to publicly admit that they screwed up with their very basic engineering calculations back in 2014 ?
and the stuff works after all.. sort of..
just not for the stated purpose (room scale or something close)
brubble
Clearly went to the wrong sweatshop.
Tammy
Google is paying to people of each age 99 dollars/hr to complete few services from a laptop … Work only for few period of time daily and enjoy greater time with your relatives .. Any one can do this opportunity!!this Sunday I purchased a great new Fiat Multipla after I been making $21683 last month .it’s certainly nicest-job however you could now not forgive yourself if you do not test it.!sh942a:=>=>> http://GoogleCityEarnMoneyOnline/get/cash/$99/hr… ♥o♥♥n♥♥♥u♥x♥r♥♥y♥v♥♥♥r♥c♥♥♥w♥♥p♥g♥♥♥r♥♥♥s♥♥♥h♥x♥♥h♥a♥♥♥m♥r♥♥♥z♥♥e♥v♥♥♥d♥♥j:::!wf183o:uwudchs
Zachary Scott Dickerson
The body tracking looks promising, and I do have some issues now and then with line of site on Vive, when laying prone for instance. But I still can’t find any games that use my trackers a year later.
Andrew McEvoy
I read on the latest update on kickstarter that they plan to demo at GDC. They had better hire some ex SAS for protection.
awilko
I wasn’t going to say it, but yeah… their GDC stall cost over $200k? You would think that money would have been better put into solving the delays, maybe faster carrier pigeons?
Mei Ling
Too little too late.
Get Schwifty!
How so? One could argue that point on a lot of technologies initially against the “mainstays”… never discount a small company doing it better.
awilko
I’m one of the original backers. It’s taken sixense 4 years to get from the “home stretch” to the “final phase” so I’m not holding my breath over this announcement. It looks like it is a bit of pre-GDC positioning. It’s a shame non kickstarter backers and potential partners / investors can’t see the comments section of their kickstarter campaign, at this stage it is pretty much 2000 disgruntled people who just want a refund.
Sponge Bob
better learn basic physics, dude
electromagnetism at the very least
I knew this was DOA from the get-go
awilko
Heh. I normally don’t like to be rude on the internet, but you might want to pull your head in a wee bit here, maybe go freshen up on the physics yourself, because you’re quite wrong and provably so.
I owned a razer hydra for years before oculus DK1 was a thing. It is an earlier version of the exact same EMF tech, licensed by razer from sixense and it works fine, it has about a 6 foot range on a much smaller emitter.
However razer took several shortcuts to keep production cost down on the hydra, like not making the handsets wireless, not using concentric pickup coils so they lacked precision, locking them all to a single frequency so you could not use them near each other, and not accounting for the spherical nature of the EMF field in software (I had to correct fot that in code myself).
You could see from all the “not in this version” notes in the hydra SDK that sixense had a much better product in mind right from the beginning, the potential of what a more expensive version could have been was obvious. That is exactly what sixense stem was supposed to be – an upgraded version of the already proven and functional razer hydra, only wireless, with a bigger emitter for better range &/or precision, multiple frequencies, higher frequency and better signal modulation to reduce latency, plus the modular additional trackers.
Basically everything that hydra owners wanted and more, by the same company who invented the technology, which we all knew works well.
Sponge Bob
Dude
the original tech behind Hydra comes from Polhemius
magnetic field attenuation will not allow wireless battery operated version because of power consumption alone
tech does not practically scale to large distances – e.g. room scale
precision drops sharply with increasing distance
This is the real reason behind their failure, not some manufacturing difficulties
awilko
First, thank you for making me aware of Polhemus, but they’re not exactly consumer grade, are they? This is confusing as Polhemus didn’t license their patented tech to razer for the hydra, that was sixense, so I had a quick google, turns out Igor Khalfin is the inventor, the patent was originally held by Polhemus, but Igor Khalfin also happens to be the guy in that sixense video explaining how the STEM improves on the hydra.
Second, can you explain to me what part of the sensors require power? As I understand it, the biggest power draw would be transmitting the signal over wifi. The wireless sensors just pick up a tiny induced current in the coils from the field. The transmitter which requires power to produce the field is not wireless.
I was under the impression the induced current was similar in principle to a crystal radio which requires no power whatsoever. Here’s a schematic of the hydra handset which seems to support that impression: https://janoc.rd-h.com/wp-content/uploads/2012/09/coils.pdf
Now I’m not an electronics engineer but to a layman that looks like it has a hard limit of around 700mA total since the the 3 AD8656s CMOS amplifiers are only rated to 220mA each on the load side. Of course for wireless you’re going to need the ADC in the trackers not the base, but still, I’m not seeing any major power draws. Am I missing something?
I’m only speculating here, but I think a far better explanation for their failure to deliver is that they simply don’t care about the kickstarter backers. They’ve already spent that money, their focus seems to be partnering up with other companies who will give them more money, the $600k they took from backers is not a lot compared to what they’ve taken from other VC investors over the years since then. Clearly it is working for them, someone must have given them at least another 200k recently for their huge GDC stand.
Sponge Bob
Magnetic field attenuates as inverse 6th power of distance from the source (relatively small coil)
Remember that the Earth magnetic field is generated by a huge coil so to speak
With small coil and not enough current going through it you can expect the strength of generated magnetic field be less than Earth magnetic field at some relatively short distance (e.g. 15-20 feet) making it impossible task to detect and distinguish
I am not even talking about parasitic magnetic fields coming from any electrical appliance nearby
This tech can only work in close proximity – a few feet at best
Magic Leap uses this same tech for hand held controllers but it remains to be see how good and precise are those
We’ll see later this years
awilko
And yet the hydra, with transmitter coils around 1/3rd the diameter of the STEM works perfectly well at a radial distance of 5-6 feet – the limiting factor is the cord length which was no doubt chosen to be just inside the operational limit of the field. Sixense claim the STEM has a radial range of 8 feet with a transmitter 3 times the size. (that said they only claim the hydra works up to 4 feet)
If you had used a hydra you would know that the distortion issue is not typically with nearby electrical equipment, perhaps if the transmitter were sitting on an electric motor, but that is not a typical use case. Rather, the field distortion you see in practice is minor and comes from things like desks with steel tube framing. That distortion is typically constant and can be accounted for in your calibration routine in software, which you have to do anyway, regardless of tracking method.
What you’re saying re Earths magnetic field also does not make a lot of sense to me. Consider the accelerometer in an IMU, the Earth’s gravity exerts a constant force, which the IMU treats as a unit length vector, this does not prevent the accelerometer from detecting small changes, since reading the delta negates the constant.
That proportionality actually was the death of a little project I was thinking about that uses very small electromagnets, as the force you can generate in a volume is directly proportional to that volume, so the relationship between 1/d^3 vs 1/d^2 meant the idea could not effectively be miniaturised. I would be delighted to be proven wrong about that. Where exactly do you get this 1/d^6 for flux attenuation from?
Sponge Bob
1/d^6 is actually power attenuation (squared magnitude)
the magnitude attenuation in the far field would be 1/d^3
makes huge difference compared to e.g. propagating electromagnetic waves (RF) attenuation which is 1/d^2 in power
I never said Hydra does not work
but tracking accuracy drops sharply as distance increases
read official data sheets for Polhemus tech – you will see it right there
I think 8 feet is near detection threshold and they need at least that much to have practical body tracking – thus the excuses and delays
room scale ? – forget about it
awilko
OK, I’m starting to see the misunderstanding.
The STEM kickstarter was 2013, the only consumer grade positional tracking that existed at that time was when people like myself strapped a hydra to their DK1. It wasn’t until a year later that the DK2 came out which locked you into a limited area and orientation in front of a single camera. At that point a DK1 plus hydra was still the best tracking available.
Nobody claimed the STEM was “room scale” when the term “room scale” wasn’t even in the public vocabulary until a year later, something that happened after the limits of DK2 tracking were apparent.
There was a video, I think it is the one I mentioned before where Igor explains the difference between hydra and stem, but I’m not going to watch it again now to verify that. He talks very openly and specifically about the range. It was supposed to have two separate modes, one with high precision and smaller area, still better on both accounts than the hydra if I recall correctly, and another mode for larger area with less precision.
Worth mentioning too – the same limits and solutions apply to camera based and lighthouse tracking, neither of those achieve room scale with a single tracker / emitter. If the STEM actually got made when it was supposed to – or even a year or two after it was due, it would have been just as easy to increase the tracked area (if you needed that – i certainly don’t) by using multiple base stations as has been done with both rift and vive.
As it is now, if the STEM does come out this year (I really doubt it will) it is completely useless to me. I already have 3 other positional tracking systems that are properly integrated with their respective HMDs. The STEM would have been awesome if it came out alongside the DK2 mid 2014 when it was supposed to, with the same specs they had working in the prototypes.
But it’s now 4 years later, so of course it is obsolete, so is every other piece of VR hardware from 2014.
Sponge Bob
you are still confused
magnetic tracking has hard physical distance limitations
optical tracking – none of that
the only difference between DK2 with one camera and CV1 with 2 cameras is that with 2 cameras (or 2 light houses) you can do triangulation instead of pose estimation
good tech is never obsolete, its just not suited for room scale tracking, not for cheap home setup at least
make coils bigger and current stronger – then it will work
but who will buy it just for removing LOS limitation ?
awilko
What you describe is exactly what the stem was sold as – a hydra with bigger, stronger coils for increased range, (plus wireless handsets and additional trackers.)
I’m not confused at all. The problem I have is that your initial statement is completely false.
You say you “knew it was DOA” in 2013 – a full two years before the vive dev kit came out and developers starting thinking about developing at room scale.
When the STEM was announced, and even a year after it was due to be released it was still suitable for 100% of use cases. So you are lying about that.
You say the handsets require too much power, I showed you the schematics. 5 watts would slag the electronics in the hydra controllers. The controllers have always been extremely low power. So you are wrong about that too.
Camera and lighthouse are both distance limited, attenuation of the laser and pixel resolution of the camera. Both can be range extended by adding additional cameras and lighthouses. STEM is no exception here. See Polhemus for working examples of multiple bases for multiple room tracking. STEM on the other hand does not support multiple bases – but only because it does not even support a single base yet on account of the fact the STEM still doesn’t exist.
You say that it cant work, and I should learn basic physics – I know magnetic field attenuation is not a simple formula. I show an example of 1/d^2, you mention 1/d^6 and go on show examples of 1/d^3. You are either misled yourself or being deliberately misleading.
Since the hydra was already adequate for tracking, from where I sit you are basically wrong in everything you say.
I was gutted when sixense started talking about adding IMUs to the STEM, that’s when I knew it was vapour, but that only happened after the vive came out and sixense started playing catchup, because now “room scale” was a thing.
Then there was their fully mobile tracking device which they offered backers. Instead of fulfilling their kickstarter obligation in a timely manner, they had gone off on a tangent and designed an entirely new device, for a completely different use case because oculus and HTC beat them to market.
I think most of the backers declined the new device, out of a complete lack of faith in sixense. Now *that* device suffers from the power issues you are talking about, but the original STEM does not have those problems, it’s issues were completely the opposite – it failed FCC emissions testing because the field from the base station was too strong.
I maintain the problem with sixense has nothing to do with physics – it can’t be impossible when they have working demo units and an existing successful retail product.
I don’t know exactly what Sixense’s problem is, but must have more to do with greed, poor financial management, lack of experience in product development, no sense of responsibility to their bakers or some combination thereof than anything related to issues with the design itself.
The hardware is perfectly capable of doing what they claimed it did, and it was from day one.
The one thing we can at least agree on is nobody would buy a STEM in 2018.
Sponge Bob
“it failed FCC emissions testing because the field from the base station was too strong”
“I maintain the problem with sixense has nothing to do with physics”
Dude, these two sentences of yours are in complete contradiction
why would they make the field too strong in the first place ?
Your kickstarter pledges are peanuts for those SV dudes –
they just don’t want to publicly admit their failure and return your money – they do have more than enough money from other investors mind you
you do not understand the difference between e.g. optical and magnetic tracking
there is no laser attenuation (unless you are in fog or smoke)
it has to do with size of detectors and their sensitivity and desired update rate of course
you lack some physics background – it is 1/d^6 magnetic field magnitude (in Teslas) attenuation in the far field
very sharp attenuation indeed limiting the distance to few feet at best. Period. Dead end . No practical way out.
awilko
Define “a few”
NooYawker
You just described 99.99% of technical kickstarter campaigns.
awilko
I haven’t backed a whole lot, a few successes and a couple of non starters. Sixense is the only one I backed where they took everyone’s money and never delivered.
NooYawker
I got burned by znaps. Simple concept. Similar product are actuallly out but they took the money and ran. Campaigns outside of new products for tech, things are much better. It takes 10’s even 100’s of millions to build a piece of new tech.
mpisc192
I think it should be different expectations backing a few dreamers with a great idea and an established company with a history of successfully made products not requiring crowdsourcing.
Konchu
I’m sure those that backed this will be happy its not a total wash. Have to see if its good though.
sipon
Overall sure we’ll ever use it , but i’ll be happy to expose this as a trophy in my office !
This is what happens to kickstarter campaigns that don’t get picked up by billion dollar companies. This could easily have been Oculus.
killdozer
Not really, Oculus delivered kickstarter project
NooYawker
What? Oculus never released a consumer product before FB bought them for 2 billion. Like all kickstart campaigns, they had demos, but no product.
mpisc192
DK1 and DK2
NooYawker
Prototypes or developer models not commercial product. Is the Hololens a commercial product?
Oculus had lightning strike twice actually, Carmack came in, invested and programmed the DK1.
Did you see what the product was before Carmack? Literally parts put together by duct tape. So let’s no kid ourselves. this is not a knock against oculus.
Sponge Bob
DK2 is true VR headset with 6dof full positional tracking while DK1 is not with its 3dof orientation tracking
DK2 was not possible without major investment far surpassing kickstarter funding
duct tape does not matter as far programming goes
and carmack is just a programmer not an engineer
Michael Gardner
The Kickstarter only promised DK1s which were delivered.
mpisc192
They aren’t a startup though, they’re the maker of the Razer Hydra and other software.
RogWilco
At this point it’s too late. What on earth were they doing the last 4 years. I’d rather have my $350 back over a device made redundant more than 2 years ago.
Laurence Nairne
$350?! Jeez, you got shafted!
mpisc192
I was a preorder and not a kickstarter…. they finally gave me a refund after a year and a half of hassling them. At first they only offered a 70% refund but eventually I was able to get a full refund.
Michael Gardner
They should be trying to license this tech to major players so it can be added in for tracking redundancy to protect against occlusion.
