4DOF Driver Training Simulator

A little something I've been working on in my shed for a couple of years now. This driver training simulator gizmo is the forerunner to a system to knock 20% off the road toll, around the planet. Once I get it properly sorted and out of hobby stage. Details at https://www.vrdriversim.com.au/moki/. A full hydraulic setup. It all operates fine, in manual mode now. Variable resistors on all motions for position detection and close to setting up an Arduino Mega as the local controller to take inputs from the control points (brakes, steering, accel) and control proportional hyd valves.

This simulator will need fully purpose developed software, in time. However, at prototype stage, I need to prove the concept which is best done via hooking the simulator up to games. Which is what got me looking around for the smartest way to do that, and here we are, popped up at XSimulator!

I had intended to just strip a proprietary game controller and distribute it on the simulator, so eg when you turn the steering wheel both the hyd activates and the game gets the input via usb, plus horn, gear position etc direct wired. Figured any scaling needed if there was a mismatch could be done in the Arduino. But it seems it won't be that simple and XSimulator needs to be used to do this? I am basically building a glorified game controller, so why not literally make it that, I figured. Is this simple logic fatally flawed?

Current prototype is a 4DOF (plus under-seat vibrator), but I have a much simplified structure and way to make a 6DOF version on next prototype.

Lateral rotation is likely to be a necessary DOF to add, plus lateral acceleration; technically that is 7DOF, but economically doable, which is my principle criteria to ensure the finished gadget is cheap to access.

I would rather have written this info all in one post, but the forum rules incentivise splitting individual elements. Sorry if my doing so has irked anyone, but do know that makes two of us!

Thank you for the explanation re process, but i thought the need to 'score coins' to have a look at the plugins did incentivise multiple non-spam posts; I stand corrected.

So, on to the important stuff: Is the simple idea of stripping a proprietary game controller and distributing it on the simulator fatally flawed? I have no qualms pulling an older Logitech one I have into tiny bits and extending wires to suit the hardware so the PC sees the same input as it would if the controller was bolted to a desk. Except it is in a sim-rig that moves simultaneously to the same control input. Could this work or is it just nonsense for some obvious reason? Thank you.

The particular game controller I was intending to sacrifice is
the Logitech formula feedback wheel E-UK12, if it is worth doing. I do also have a decent Thrustmaster I would be more reluctant to kill as it is my 'reference point' to compare game response etc. I do have an Arduino Mega and lots of shields 'to go'...

Hi to a fellow Novacastrian. All the best for your project - looks like a beast . Agree with @yobuddy, why re-invent the wheel - Simtools + arduino/feedback system and a interface circuit will easily do what you want. Link to my project thread if you are interested.https://www.xsimulator.net/community/threads/seattimes-sims.11757/

I really appreciate these comments, thank you. And I love that ppl are doing this kind of stuff; I thought the interest / capacity to do this kind of 'out there' tinkering was a dying art, given the trends in modern life. But here lots of you 'know how to think and turn it into action' miscellaneous nutters are. Long live the nutters! Who, as any of us know, are at least an order of magnitude more 'into it' than regular nerds - who themselves can be awesome (on their good days...).

As it happens Oliver, a final year Mech Eng uni student, has just started doing some modeling of the simulator with a view to develop the control algorithms, and whatever else that will ensure he gets a good mark in his thesis! He has all my Arduino gear atm and the intention is to validate the modelling against the real world test rig. For my part, I need to do some hackwork like get all the field devices wiring pulled back to the control box, as well as sort out exactly how I am going to do force feedback to the steering of the many possible ways. I also think I should simply stump up and support this sort of project and purchase the SimTools lifetime package because that is the fair thing. After Oliver has the chance to compare the other couple of main alternates so he can trust this old dog's judgement call!

I have a question, but don't want to be one of those ppl that asks questions first before doing the minimum 20+ hrs of reading that this topic needs to get to have a reasonable grasp on the issues / techniques. With all of ~3hrs under my belt so far, in the interest of efficiency / better focus, can I ask if, for the sort of project you can see this is, does adding a Thanos AMC1280USB motion controller to the mix get useful value / flexibility / capability that I am not going to get with SimTools and Arduino alone? Thank you.

I use two 1280s to drive my rig and they do the job quite well, but for what you are doing, it's likely that some custom code in a Arduino, or similar controller would work the best. May also be a good idea to look at one these controllers - ESP32.

I am 'fully loaded' with Arduino stuff, to be sure to be sure... Didn't know exactly what I would need, so everything that sparkled wound up in the grab-bag of stuff. I am really looking forward to 'getting into' that bit, but MUST do some other stuff first (with the aim of generating a bucket load of funds to take this thing to full scale, rather than need to look at crowd funding or investors, the latter I really don't want to do!).

I have potentiometer based manual control now and it is fine. Turn and tilt is via servo proportional valves that just need a +/-10V signal. The trick this needs to be good at is eg turning a bit left, then a bit more, then a bit less which must be seamless and, with these valves, it is. The bounce (heave) valves are on-off cheapies that can take 4Hz relay signals so I figure I can 'flutter' them to get what I want, or upgrade the valves to basic proportionals if needed, no way the servo ones are warranted there (about $4k new, reconditioned ones 'on loan' from a friend, while I get this thing sorted). I have a vibrator motor under the seat, a 110W proper industrial unit; currently using a 4-20mA manual controller but have a shield to run that too.

What I was originally thinking, flawed, was that I will calibrate the sim to move 'about the right amount' from direct inputs, so no possibility of lag. I may have a scaling table set up for different games / software, but basically relying on matching them. This is why I thought stripping a game controller and mounting it so same control signals were sent simultaneously (but I didn't want to reverse engineer the way a game controller generated its signals so they could stay separate, just mechanically locked together). But the whole thrust of what is being said in these forums is that synchronisation is not an issue, so just use the game outputs to control sim movements and have a sorted 'plugin' to provide appropriate scaling factors. Fair enough!

Just for 'calibration purposes', the kind of performance I CURRENTLY have (can tweak this with pressure changes) is:
Roll slew rate avg 0.25rad/s right and 0.41rad/s left through +/- 30 deg.
Pitch full stroke avg 0.5rad/sec accelerate and 0.89rad/sec brake through +30 / -35 deg.
Heave 16cm/s through 20cm. This is limited to that by flow control valves, could easily adjust for more / less and different rise & fall.
I haven’t attempted to measure acceleration rates, but they are pretty quick!

I may add a rotational vector to help with position on the road sensing. Will experiment with this first as not sure if need it and anyway the second prototype design is much easier to add that to, as well as a direct linear acceleration vector, making it 7DOF