1. Do you need support for Assetto Corsa Competizione? Please use the proper forum below and ALWAYS zip and attach the WHOLE "Logs" folder in your c:\users\*youruser*\AppData\Local\AC2\Saved. The "AppData" folder is hidden by default, check "Hidden items" in your Windows view properties. If you report a crash, ALWAYS zip and attach the WHOLE "Crashes" folder in the same directory.
  2. FOR ASSETTO CORSA COMPETIZIONE: If you report issues with saved games, please always zip and attach your entire User/Documents/Assetto Corsa Competizione/Savegame folder!
  3. If your game executable is missing, please add your entire Steam directory to the exceptions in your antivirus software, then run a Steam integrity check or reinstall the game altogether.

Article: Why Force Feedback In Computer Simulators Does Not Work?

Discussion in 'Chit Chat Room' started by liakjim, Aug 21, 2017.

  1. liakjim

    liakjim Alien

    A friend of mine, shared this article http://www.taringa.net/comunidades/...Porque-el-FFB-no-funciona-en-Simuladores.html
    @Lord Kunos i would like some thoughts if you have some time to spare. thanks

    There is an translation in English right below the text in Spanish :

    Why Force Feedback In Computer Simulators Does Not Work?

    Over the last decade we have seen dramatic improvements in PC hardware power, quality of video presentation, number and accuracy of simulation models and yet one thing still fails to impress. That’s right, force feedback, or in short, FFB device. As a consumer product it has been around since mid 80‘s and greatly improved along the way. Steering wheels, joysticks, aircraft yokes - they all seem to push, pull and vibrate but still don’t come close to feeling like controls in actual car or airplane. So what is going on here? Is this a problem of design, manufacturing or the whole concept? I have designed and built a number of FFB systems completely from scratch - both mechanically and electronically. These steering wheels are 10 times stronger and faster then consumer ones. These systems have low inertia, low friction and two orders of magnitude higher force control accuracy than what came before them. They can respond to force commands and send position data to PC up to 1000 times a second. So do these advanced wheels feel 10 times more real then the consumer wheel when plugged into a typical PC simulator? No. They are more powerful, fast and smooth but no more real... Maybe the problem of realism lies somewhere else but not in helical gears, belt drives and precision torque control? Maybe somewhere along the whole concept of simulation there is something that just does not work is it should? I will try to get down to the very basics of how we model the physical objects in virtual world and then interact with them from our real world. It is easier to work from an example then from formulas, so... Let’s consider a simple road car simulator. Road car - because its dynamic modelling is easier than aircraft or submarine. Simple - because we would strip everything nonessential from it leaving just bare minimum - just enough to understand the concept. It won’t need to travel fast so we take out the engine. It won’t need to slow down quickly so we remove the brakes. We will not slide into tight corners so we leave just one steerable front wheel. Since the cornering forces are low, even steering rack would be an overkill so we will leave it out too. We have one and only push trike simulator! Apollo has been sent to the Moon with a guidance computer that had 2 Kbytes memory and 2 MHz CPU speed. Surely a modern computer with 4 GBytes memory, four cores running at 3 GHz each and a steering wheel with special profile helical gears and dual motors can simulate this trike so well that were you to close your eyes you’d feel like you are 3 years old again? In reality our imaginary simulator feels nothing like real trike. Why? Let’s start from the beginning and look at how our push trike simulator has been written. The way most (or maybe all?) simulators are. We have created precise mechanical model of the frame and all other important parts. We calculated or measured all the fixed to the frame and and moving masses, their moments of inertia, locations of their centres of gravity. We have placed constraints on the way these parts can move, rotate or slide with relation to each other. We have written down formulas tying together acting forces and movement of these parts including friction, elastic compliance and viscous damping. We have measured aerodynamic drag, and even lateral forces acting on each part - including the points where these forces are centred. We have run the tyres on a drum machine and plotted cornering force and aligning torques at all imaginable loads. We run the simulator and test drive it. It still feels nowhere like a real thing. Why?! Let’s see how we handle force feedback in our simulator. Any FFB wheel works in a very simple way. It receives force demands from the simulator running on a PC and sends back reports with accurate position data to let simulator know what way and how much we have turned it. It sounds obvious but let’s point this out again: simulator sends force commands to the wheel and receives position data back from it. How do we decide what force to sent to the FFB wheel from our sim? We find the grand total of all forces acting on the steering axis based on front tyre vertical load, rotation speed, position angle. How do we know the front tyre position angle in relation to the vehicle? Simple! We get it from the steering wheel, right? We don’t even have to divide it by the steering rack reduction ratio since we don’t have a steering rack. Let’s step back and repeat this again. We send the forces acting on a car front wheels to the FFB steering wheel and moments later we receive back new position of our car front wheels. Doesn’t this strike you as strange? Maybe not, so let’s focus on this. We have offloaded the whole process of simulation of the front wheels and steering rack dynamics onto the FFB steering wheel. What should have been precision simulation of acceleration, damping, friction of the front wheels steering movement is now outsourced to a device with a few plastic cogs and less then 7 bit force accuracy.

    [​IMG]


    But even the best steering wheel is not designed to simulate anything. It just moves when you ask it to. If you hold it tight, it does not move. That’s all. You might argue that wheel has some natural moment of inertia and damping and they can be considered scaled down properties of the front wheels. No, the moment of inertia of any FFB steering wheel is less than even our simple trike’s steering and front wheel moment of inertia - let alone the inertia of front wheels and steering rack of actual car even reduced by steering rack gearing. OK, moment of inertia simulation is pretty obvious - we have two massive front wheels on any car and they need some good effort to make them accelerate. They have “weight.” But why damping? Cars don’t have dampers in steering systems. Sure, they don’t but requirement for damping simulation comes from natural damping in steering rack and existence of gyroscopic couple - the result of forces acting on a front wheel spinning around horizontal axis and contact patch moving towards one side of the wheel while cornering or just compliant suspension (variable camber.) The resulting counterforce is proportional to the speed of steering wheel rotation. Which in the end is equivalent to damping. Gyroscopic couple damping effect is proportional to the vehicle travelling speed and is one of the reasons why steering “stiffens” with higher speeds and motorcycles remain stable at high speeds. But we have digressed. We have concluded that maybe if steering wheel has some damping and inertia it can simulate the front wheels for us? At best simulation of damping in FFB wheels is mediocre. Even worse, no consumer wheel I know of can simulate inertia. This is due to the fact that it requires very accurate angular acceleration estimation. This is just not possible with simple encoder even with few thousands counts per wheel turn. And now is the worst bit. As soon as you gently place your hands on the wheel its damping and inertia increase several times. You grab the wheel tighter and suddenly the car has ten times more massive front tyres. No car in the world has tyres with variable moment of inertia! So what would be the “right” way of getting around this? How can we simulate the car front wheels inside the sim? It won’t be a problem if we had all the necessary input data - all the forces acting on the wheels, their position, moments of inertia, etc - and wouldn’t let anybody else interfere with the accurate simulation. Simulated by PC software “Simulated” by FFB wheel Remember that we have noted earlier that FFB steering wheel traditionally receives the force demands from the sim and sends back the wheel position? One way of solving our simulation problem would be to turn it all on its head and instead send position demand to the FFB wheel and receive back the force applied by driver. From first glance it does not make sense. But look how the real car works. It presents you the steering wheel in a certain position and then you apply force to it in a hope that it will move where you want it to be. Then car then moves the steering wheel through its tyres according to physics laws. The wheel is happen to be in a certain position and you apply forces to it to make it change this position. You control the car by applying forces and not by placing the wheel at a certain angle. Think about it. You are turning a corner and going a bit wide. What do you do? Do you add extra 10 degrees to steering? No, you add a little force and the wheel turns a bit more. How much? You probably have no idea and neither do I. What if your street car had a wheel that had no force on it when you turn it? In other words you would control it by the position of the wheel. You probably wouldn’t be able to drive it. But it would be perfect for any simulator to model because this is how simulators are written to work.

    [​IMG]


    So once again, modern car presents the steering wheel to the driver and driver applies force to it to make the car change direction. This is completely the opposite as how FFB devices made to operate. Can we build such a device? Should it then be called Position Feedback? PFB wheel? If you are familiar with industrial automation and controls then you probably already guessed that ideal system for our sim steering is a servo system with stiff position control where torque demand from PID loop not only applied to servo motor but also sent back to the sim as force applied by the human driver. Our car sim physics update loop works in the following way. The software takes the front wheels position, adjusts it according to steering rack configuration - in other words scales it down by gearing ratio - and sends it to the steering system. This now becomes as servo system new holding setpoint. External force applied by the driver is measured and sent back to the sim. The sim combines this steering rack force with other known forces acting on front tyres and then calculates wheels vertical rotation dynamics - angular acceleration based on wheels moment of inertia, rotational speed, and finally arrives at new wheels position. Which is in turn sent to our steering device, becomes a new setpoint, new steering force duly arrives back and everything starts over again. Perfect harmony. Our push trike sim finally feels as good as a real one. Can this be done with an ordinary FFB wheel? Yes, to some extent. If you set up a very heavy spring effect (P component) accompanied by critical damping (D component) then you have some sort of a rudimentary servo system. Getting the motor torque out and into PC is more problematic as you can only guess it via displacement from a setpoint (P-component input.)

    [​IMG]


    What is the morale of the story? The simulation software programmers rely on hardware manufacturers for being part of their system while hardware manufacturers never expected to be “in the loop” at all. Few people in the industry understand how “the other side” or even overall system works and there is no available information or discussions on the subject. I hope someday this will change for the better! Leo Bodnar, July 2011 Simulated by PC software FFB wheel is just input/output
     

  2. Similar Threads
    Forum Title Date
    ACC Gameplay Dust particles too small Jan 20, 2019
    ACC Graphics - Tracks and Cars Please add some more particles effects Jan 16, 2019
    Suggestions Air Particles Jul 15, 2017
    Suggestions Smoke, Particles, and Broken Glass Really Need Improving Oct 8, 2016
    Chit Chat Room two questions, dynamic mirrors and particles on the air. Jul 22, 2016
    Chit Chat Room Interesting Article - Fake Engine Noise Jun 14, 2016
    Chit Chat Room Great article "Assetto Corsa: Are PS4 and Xbox One ready for a true driving simulator?" Jan 22, 2016
    Chit Chat Room Interesting article : how top-tier race cars compare Aug 13, 2015
    Tracks modding Remove dust particle from track surface, but keep mud chunks, possible? May 30, 2015
    Chit Chat Room Interesting article with formula renault 2.0 driver severin austerschmidt May 6, 2015
    Physics modding - Cars & Tyres How to change dirt/dust particle color in 1.1? Mar 16, 2015
    General modding discussion New particles grass texture Sep 11, 2014
    Tracks modding Solved: blender particle system -> screwed up trees in ac editor Aug 8, 2014
    ACC Controls & Peripherals G923 Force Feedback causing crashes Sep 15, 2020
    Controllers and Peripherals Bugs & Issues G923 Force Feedback causing crashes Sep 15, 2020

  3. Stereo

    Stereo Alien

    This is why when people say "I can judge physics completely without an FFB wheel" that's just crazy to me, the force wheel is inside the physics loop, at least in sims that have it coded to perform that way. It's not info you see on the screen or hear, it's a completely unique set of physics dynamics that changes how the car works.

    For example an FFB wheel can self-steer and correct slides; if you just have a bland control wheel with 0 force of any kind, that physical behaviour that happens in a real car won't happen and it's just not really simulating anymore.
     
    Seria17hri11er and aphidgod like this.
  4. Article is 6 years old...
     
    mms likes this.
  5. liakjim

    liakjim Alien

    So?

    Στάλθηκε από το m2 note μου χρησιμοποιώντας Tapatalk
     
  6. Valid points, but a lot has changed since then
     
  7. mms

    mms Alien

    Here's the original thread, there's some interesting comments there too:
    http://www.virtualr.net/why-ffb-in-simulations-does-not-work-by-leo-bodnar

    I think Leo Bodnar exaggerated a bit when saying FFB does not work in sims, IMO it can work pretty well as we see it in most sims today, however it could be better if the interaction between the software-hardware (API) would model RL differently based on how things really happen. Unfortunately it would take too much effort to change this today.
     
  8. michael_

    michael_ Simracer

    It may be crazy to you, but my brain tells me that the hands I see in VR are mine. They are clearly not.

    My brain tells me I'm moving when I'm clearly not - and that movement is related to the dynamics of the vehicles.

    i.e when I move in TF2 after playing a lot of assetto corsa I note a massive difference in my "handling"

    If I bump into a cloaked spy in TF2 and my character is knocked - I "feel" it - there's nothing to feel - there's no FFB on the WASD keys or the mouse, but
    I still feel it and so I turn and start looking around for the spy I "bumped into"

    If I get lag in a game and don't get 60fps I feel that as I turn. If my mouse/mousepad isn't clean and it doesn't track, I feel that.

    So, without a doubt people can judge the physics without a FFB wheel - all they need are eyes and a brain. If they've got VR, well, their brain will happily tell them they are sat in a car that's moving along. Far stronger effect and "crazier" than anything a FFB wheel is telling you.

    Crazy? Well, as Michael Abrash said recently :-

    The reality we experience is constructed in our minds, based on a great many assumptions built into our genes and learned during our lifetimes, together with the very sparse data that comes in through our senses.

    All Reality Is Virtual.
    That’s a strong statement, and it’s not obvious if you haven’t thought about it before, so I’ll say it again—the reality we experience is a construct in our minds, based on highly incomplete data. It generally matches the real world well, which isn’t surprising, evolutionarily speaking, but it’s not a literal reflection of reality—it’s just an inference of the most probable state of the world, given what we know at any one time.


    The reality we experience is whatever our minds infer it to be based on perceptual inputs, regardless of the source. So if VR can provide the right perceptual inputs, we can have whatever experiences we want, and those experiences will feel real—they’ll be genuine experiences.

    Worth reading the whole blog : https://www.oculus.com/blog/vrs-grand-challenge-michael-abrash-on-the-future-of-human-interaction/
     
  9. Stereo

    Stereo Alien

    I'm not saying you can't get a sense of what's happening physically, just that it's an incomplete picture without the wheel, and not only that, but the physics are fundamentally not working the same way without it.


    I don't know if I've read the article before but I've definitely had thoughts along the same lines, that instead of basing forces on having 2 force directions, which you send to the wheel, you would say where the 'simulated' steering column's point currently is, and only use a (relatively strong) 'centering' spring in the wheel to make it try to reach that spot, and treat the difference as the user inputs. This would for one thing allow the wheel hardware to operate in a tighter loop (a wheel's onboard hardware, no usb, can simulate a spring force of higher fidelity, and layer critical damping onto it so the wheel doesn't oscillate around that point) and for another let the game "dyno" the wheel itself, by moving the centerpoint around and seeing how fast the current wheel settings follow, which could theoretically then be used inside the simulator to set up the responsiveness of the wheel.
     
    Last edited: Aug 22, 2017
    Seria17hri11er likes this.
  10. ouvert

    ouvert Alien

    Nice article, sure, FFB doesn`t feel like real car but do we really want to have real feeling in sterring wheel? in sims we are getting more info through wheel than IRL while missing some other informations and our brains can sort off addapt to those differences, link what we feel in FFB to what we feel in real car (through steering, body, ..) and trick us to have realtively good experience even on consumer wheels. Having just real life feeling would feel like you are missing something I guess ..
     
    LATE4APEX, dajdosta and Dean Ogurek like this.
  11. Dean Ogurek

    Dean Ogurek Alien

    Exactly, FFB helps us fill in some of the feedback gaps missing from the virtual world; I very much prefer that to what real steering feedback would feel like in Simulation. That's not to say FFB methods couldn't be improved but, I'd say what we have now is pretty great and even more so if you use a DD-wheel; of course, the Bodnar system is the cream of the crop. :)
     
    Haiden773 and dajdosta like this.
  12. paul thomas

    paul thomas Racer

  13. paul thomas

    paul thomas Racer

    I was always under the impression that AC 's FFB was pretty pure as it comes from from the forces thru the steer column. If you want extra effects you have to turn them on via the launcher.
    Dean
    I'd also wouldn't be surprised that he bodnar system isn't the cream of the crop any more. A simucube with a 4 million encoder and a kolenmorgan will easily feel as good or better than a Bodnar . Add the new wip firmware into the mix and it will blow it out of the water. Of course all Lee would have to do is develop his own reconstruction filter and it will be game on again.
     
  14. Andrew_WOT

    Andrew_WOT Alien

    I do not think this is accurate, from my understanding:
    a) all ffb effects come from physics engine, but not necessarily from steering column, mostly suspension and tires
    b) extra effects are just amplifiers that enhance already existing signal, the only exception is curb effect that is present even on flat curbs.
     
  15. kunos

    kunos Alien

    in AC, "pure ff" is:

    force and torques from tyres/springs/dampers/arbs -> suspension geometry -> steering rods -> ff output

    "Road" effects
    Car body vertical G -> ff

    "Slip" effects
    Tyre slips (all 4 of them) -> ff vibration

    "Kerb curb whatever" effects
    IF FRONT_LEFT IS ON CURB OR FRONT_RIGHT IS ON CURB THEN VIBRATE THE **** OUT OF FF

    "Understeer" thingy
    I won't acknowledge its existence, it's all a government/illuminati conspiracy

    From 1.15
    "ABS" effects
    ABS on -> ff vibration

    Whatever comes from the stuff above is then send to the "FF Post Process" system that will either leave it intact or pass it through the gamma or LUT provided.

    I remember the article when it came out I didn't read it again today but I remember thinking it was pretty correct. This is super obvious when working on motorbike simulations where the front "steer" is such an essential part of the system and almost impossible to simulate properly because of hardware limitations. With cars the limitation is still there but it's not as bad.
     
    Last edited: Aug 22, 2017
  16. ouvert

    ouvert Alien

    yes, "pure" as comming directly from susp physics through steering, that is pretty much what all good sims are doing ... luckily for us they are represented in FFB wheel in much more suitable way for our limited situation behind computer screen/VR :) ...
     
  17. Berniyh

    Berniyh Alien

    Short question regarding this:
    So if you drive e.g. a 962C which has huge tyres, stiff springs and lots of downforce, shouldn't the ffb forces be much higher than let's say a 718 RS Spyder with thin wheels, wobbly chassis and no downforce whatsoever?
    Because it doesn't feel like that.
    Or is it just that in the latter case, the effects are amplified so it doesn't feel dead on low-force wheels?
    So in that case, for "pure" ffb, you would need to set a gain that rescales/downscales it to the original level?
    I guess you're talking about the slider for the curbs here, right?

    Asking, because the rattle in the wheel can be observed if that slider is at 0%, thus the effect should be off.
    However, without that slider, it seems to be a bit different than with, because without you get the effect on some curbs, but not on all curbs.
    Can be observed e.g. on Brands Hatch.
    as in ffb effect?
    Some Fanatec wheels have special vibration servos which I think respond to a rumble effect like it's present in some gamepads.
    Not sure, but I thought that's an effect separate from "normal" ffb.
    Also, they have vibration motors in the pedals, but I don't know if you can control them from external software. Might be possible though.
     
  18. ouvert

    ouvert Alien

    without that slider it is based on load of tyre currently passing through curb ... you get way less vibration on unloaded tyre ... slider makes it vibrate on every curb regardless the load
     
    kunos likes this.
  19. kunos

    kunos Alien

    obviously yes.. but cars have a "FF multiplier" in the physics data that is designed to put the FF output in a range that can then sent to hardware. This multiplier is about 8 for the 718 and 1.7 on the 962C so it's compensating for a almost 8:1 ratio of forces.

    Curbs are obviously coming through the normal "pure" FF just as they would in real life. Some people think that is "not enough", mostly kids used to arcade games where all the could feel was curbs.. so we ended up putting this terrible fake shake to make them happy... but it's totally unnecessary and fugly.
     
  20. Berniyh

    Berniyh Alien

    Could be, but there's more to it, because I really mean that on some curbs there is nothing at all and on some you do get effects.

    Have to admit I never noticed that it should be load-dependent (which I assume means that the stronger you turn in on the curb, the stronger the effect should be).

    (Everything with that slider set to 0%, of course.)
     
    vegaguy 5555 likes this.
  21. Berniyh

    Berniyh Alien

    Ok, that's really a lot, makes sense. ;)
    Will try it this evening, but I guess setting approx. one 7th of the forces will feel a bit dull, even on a direct drive.
    Ok, makes sense, so the differences in the curb behavior is because some of the curbs are modeled (in the track model) and some aren't?
    Or differently modeled?

    Edit: setting that curb slider to 100% on my DD would be a good test to check if all of the screws are still tightened up completely. xD
     
    vegaguy 5555 likes this.
  22. Differently modelled. Some kerbs are flat painted concrete sections, some have a serrated profile sending a repeating vertical load through the suspension that you will feel in the wheel.
     
    vegaguy 5555 likes this.
  23. ouvert

    ouvert Alien

    well stonger you turn in, more weight goes to the outside and less on inner wheel that is touching the curb so you should have less vibration ...

    with DD yoy might be close to IRL feeling .. with my low end consumer wheel it is nowhere near the violent shaking those more profiled curbs do to you :).. but you feel a lot of curbs through chasi instead of steering ...

    It depends on how deep you`ll go in the curb too.. profile is changing ...
    On this picture you can see inside curb on T3 (Cimini 2) at Vallelunga .. if you go deep in you can feel it pretty strong even on less loaded tyre, if you just touch it on edge not so much ..
    IMG_20170609_201139.jpg
     
    Last edited: Aug 22, 2017
    vegaguy 5555 likes this.

Share This Page

  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.
    Dismiss Notice