Building a Rig for VR ?

@Linkup posted this in another thread and I initially thought that we had been through it before, but on reflection think it is a good idea to combine everyone's idea/thoughts and tested designs of how to best work with a Simulator rig and VR in the one thread.

Now that we have an idea of when it should be arriving(April for most of us) can we start up a thread for VR specific motion rigs? I mean a top to bottom optimization of a 3DOF setup to give the best VR experience, very low latency between telemetry input to motion output, high res Pot for finer control and accuracy, lighter top frame and seat, arduino/jrk settings and high rate data, and of course how to tweak the motion profile for VR. Basically a guide to what can be adjusted or improved on own end and maybe for devs too.

I'm going to add my thoughts to this thread soon (busy for a little while now) and invite other members to do the same.

I'll likely just add to this when I have time:

Ok. Note the following is all just my Opinion - you can learn from it, or ignore it.

Decided that pictures and videos speak a thousand words, so I will talk about my two sims and how they compared in regard to using VR.

My previous Sim, it ended up having 5 axis.



This was originally designed as a standard shoulder mounted two DOF mounted onto a surge and traction loss table, so it was driven by 4 x 200w 12v DC motors using JRK motor driver cards. Another axis/motor (heave) was later added via a modification to the seat itself. This was actually a nice sim with good low latency in the upper 2DOF frame, but which got progressively worse in the lower surge and traction loss tables due to increasing inertia due to weight and the pendulum effect (high CG). You can only do so much with JRKs and 200w motors. This was still not so much an issue with a single open screen, but then I purchased a Oculus Rift DK2.



This was a step up in simulation, but as the frame was never designed for the Rift the camera mounted to the frame use to bounce around due to flex on any fast changes in direction and lose track. This was not good, so the camera was moved off rig and is how the video above was taken. Unfortunately the large surge table movements created an unwanted change of view in the Oculus which was often a bit off-putting - but you got use to it. So if you have a Two DOF with limited movement mount your camera off the rig.

I knew the rig has some shortcomings and as CV1 was in the pipeline, started work on a new rig with VR at the forefront of its design. There was also allot of talk at the time on the site of Camera tracking issues with the Rift and motion, which from my own experiments I isolated down to low and high speed camera oscillations. The Oculus headset never seem to be the issue due the natural absorption effect of a human body/neck/head. So what I needed was a frame that stopped low speed oscillation, while also absorbing high speed oscillations. From my and mainly my Sons experience with my last Rig, I also knew that I needed the rig to have a fast and large enough range of movement to trick the inner ear to reduce/eliminate motion sickness. So after a few changes in direction a 6DOF stewart platform was born:



To reduce latency, I built Carbon Fibre framed ballscrew actuators they have low weight/inertia, available long travel, while still being capable of a good speed and torque due to their high efficiency. Ie 90% compared to around 50% for most worm drive gearboxes. The Stewart platform is also by design (all triangles) very stable and rigid. To power all this I used 12V 48amp DC motors with dual Monster Moto Drivers and 40 amp power supplies for each axis.

The actual sim frame was also designed to lower the moving part of the rigs CG, therefore lowering inertia, the pendulum effect and therefore again latency (I'm current building a carbon version of this).


This gave me a stable, low inertia rigid platform which is ideal to mount a camera to and has low latency in all DOFs.
The camera is wrapped in foam sheet - I had to try a few different densities to find one that worked with my rig (to reduce the high frequencies oscillations) and fixed to a steel frame which has been welded and braced to the main frame (to reduce low frequency oscillations). I think a mistake allot of people have made is expecting the foam to reduce the low frequency oscillations, this will never happen. The foam also needs to be harder then what you may think, as you don't want the camera to be able to move and generate its own oscillations.

Myself and my Son can attest to how well this works and how it has reduced any feelings of motion sickness from my previous rig to nil. Here it is in action.





Hope this helps someone in designing a rig suitable for VR - I will add more information if it comes to mind.

Thanks for getting things started @SeatTime and @noorbeast. I've got all kinds of questions, but won't drop them until I've at least read through your respective threads. Also will be looking forward to seeing @SeatTime 's 6DOF with the Rift CV or Vive CV and how that experience is, I imagine it will be incredible!

I've already started a 2DOF so I will try to finish it before building something with all the advice here. I'll add that my getting into this stems from the idea of how amazing it would be if VR tuned motion rigs could be sitting in people's homes(especially mine!) to help with motion sickness in anything with movement and of course increase the immersiveness of their gaming. If I'm right most of the user's here will be building with VR in mind sometime this year and never look back. The DIY route seems like the only sensible route as commerical rigs have such high price tags.

I eventually want to build a rig that can simulate the random yet fine motions of someone walking/running and perhaps make first person VR experiences possible without getting sick or limiting the user's control/ability as done now with room scale and seated VR.

Once again thanks for getting this started and hopefully we can get a VR tutorial to give everyone an idea of what they need to consider.

@noorbeast You have some interesting work in the pipeline and I would be interested to know if your haptic chair has any effect in lowering motion sickness as I know motion can. As an early adopter of DK1 I expect that you are immune, although possibly you can find a suitable subject like I had with my son.

Excellent work getting a grant @noorbeast , I would look into Simtools/Harptic support via simple software on a Arduino, it should not be that difficult with what is already out there.

Part 2.
Centre of rotation: This is something that is really not talked about much, but when I designed my new rig I purposely lowered the centre of gravity not just for a mechanical advantage. When a simulator rotates, it is important that the centre of that rotation is similar to what is being simulated. I purposely set mine up to mimic the rotation of a cars suspension. The more a simulator moves the more important this will become. I'm certain this has had a impact on reducing motion sickness in my new sim. FI if proper 6DOF math is being used, it is possible to manipulate this in software, so I do not know if this will be a feature of SimTools V2?

@SeatTime Do you happen to have a close up picture of the wrapped camera? I want to start looking at tripod type holders for the camera to mount to and maybe something other than foam wrap to give that slight dampening need.

No, but it really is just a camera wrapped in foam. Ie. Isolated from the frame. Works for me, but the rest of the rig must also be suitably built as explained above, or it may not work. But it should not stop you from running your own experiments if you wish.

Suitably built indeed. I can also try testing different spots. I'm not sure where the high speed oscillations would be less or more on the rig , but I've been thinking of how to test for that. Should be plenty of camera mounts that can be used, anything with a pendulum dampening element should work or maybe even o-rings and cushioning.

if your building a full motion sim, then a pendulum camera rig will not work as the camera must follow the movement of the sim.

Hi Everyone!

I am itching to start building the a good cost affordable rig for vr but have never done anything like this before. I can weld and solder and am sure that I can build it if there is a step by step guide. Does anyone know of a good VR build (I'm guessing the 6dof set up is the best) site that will sell a kit that come with great instructions or a forum post of all the parts and steps needed? Thank you!

-Ryan

Awesome! I hope to see some good info here. This is exactly what I am looking for. I get my Rift in April!

This thing is awesome! I watched the video and looked at the Pics, but I am having a hard time seeing how it all works. What kind of motors are those? How much did this thing cost! This is my issue. I want to build something, but its a balancing act with how much I can spend! (Wife/Kids and all)

My build thread is here http://www.xsimulator.net/community/threads/seattimes-6dof-dc-build.6106/ Cost - well as you have to have 6 x everything it is never going to be cheap, some like @SilentChill (see his build thread here http://www.xsimulator.net/community/threads/6dof-project-using-12v-dc-motors-mms-and-arduinos.6948/ ) were able to acquire major components (good quality Motors/gearboxes) for free which would have greatly reduced the final cost. Not a small decision for you to make, as it is quite an investment in time and money. If you want to keep the Wife on side - IMO recommend you start with a 2DOF seat mover, as you can mount the VR camera off-rig and negate all the on-rig issues. Not in my opinion as good as a 6DOF, but way better than no motion at all.

Thanks. I found that thread just a few minutes ago! I'm so jealous. I am thinking of doing something like http://www.xsimulator.net/community/threads/2dof-foot-mount-sim-powered-by-12v-electric-jacks.7601/ or http://www.xsimulator.net/community/threads/diy-2dof-simulator.4793/page-9. Do you think a foot mover would be just as good for vr as a seat mover? I'm new to this all and don't have a benchmark for comparison. Thanks!