Search Unity

  1. Megacity Metro Demo now available. Download now.
    Dismiss Notice
  2. Unity support for visionOS is now available. Learn more in our blog post.
    Dismiss Notice

Unity 5. WheelCollider component is unusable.

Discussion in 'Physics' started by SeriouSerg, Mar 4, 2015.

?

What do you think about new WheelCollider component?

Poll closed Apr 3, 2015.
  1. It is unusable and needs reingeenering.

    59 vote(s)
    80.8%
  2. It is suitable for use

    14 vote(s)
    19.2%
  1. mkgame

    mkgame

    Joined:
    Feb 24, 2014
    Posts:
    592
    But for my purposes the wheel animation is much more present. In Unity 4 my tanks where stiff, and in Unity 5 they can move like a cadillac. From the view of a gamer, i would say, the realistic behavior isn't really needed. Mostly it is enjoyable if the physic is overdone. I have to overdone the physic for an RTS game, because the units are smaller and the player should feel/see the physic, when that physic cost me so much time, effort to pathfinding the units...
     
  2. MT3

    MT3

    Joined:
    Nov 14, 2013
    Posts:
    12
    I am at a loss as to how to proceed with Unity. After waiting for an update that made the hundreds of collider errors go away in Unity 5, I'm now faced with wheel colliders being completely wonked and joint motor completely failing to function... I can't make a career of testing unity and filing bug reports. I can't expect to produce and maintain games on an engine that only has a "previous and unsupported version" as it's only working version, nor can I create a major rewrite of my games every time Unity updates a major version.

    I would love to hear Unity comment on this situation as it affects me (and I suspect hundreds of others) in drastic fashion. At every previous major update I was able to move my developments to the new version with minimal problems. Version 5 has broken an unacceptable number of hours of work... and apparently completely broken many previous functionalities. I'm not interested in throwing blame. It's Unity's fault, it's PhysX' fault... doesn't matter at all. What does matter is investing the future of my company in a certain game engine and then finding that I have no way of knowing that the engine will perform the same from version to version.

    If we have come to the point that previous versions of Unity cannot be kept compatible with existing hardware and only games produced in a matter of months with a life expectancy of about that same amount of time...are practical, we have come to a dead end. Continuity is a major aspect of developing and maintaining an online or PC based simulation/game.

    Perhaps the only mistake here is integration of the new PhysX model before it had been fully tested and proven. I can't believe Unity knew of these problems and simply shrugged and moved forward... I must believe they were simply unaware of the extent the physics changes would negatively impact the Unity engine. Physics based games in particular need reliable and logically sound functionality to work properly. Unity seemed a sound choice for simulation until version 5.

    Please Unity make it clear for us...

    Should we simply decide if we can live with this level of unpredictability
    or
    should we hold out hope Unity will actually fix these issues and return the features we previously integrated into our developments?

    (note the skateboard that must weigh in at 1500 Kg to operate "correctly")

    I have had great faith in Unity and still believe they are a great company with a powerful vision. I hope that we haven't reached a crossroad that will divide us, but I must decide if I can afford to write my developments over and over to keep pace with changes in Unity while simultaneously adding and removing basic game features at the whim of PhysX.

    I may be alone in this conviction... but I would also appreciate feedback from other developers on this subject.

    Ever hopeful,
    M Miles

    [EDIT]
    Can Unity offer a switch to enable one PhysX or the other? This method could be employed any time Unity introduces "breaking" changes of this magnitude.
     
    Last edited: Jun 6, 2015
    Nition likes this.
  3. RLin

    RLin

    Joined:
    May 10, 2015
    Posts:
    133
    After reading the docs and other official Unity documentation on the Unity 5 wheelcollider, I found that they had a strong trend of implying, though never actually stating, that the Unity 5 wheelcollider is made only for cars. There is no documentation on using it for other vehicles; in fact, the docs suggest that it is specifically fine tuned for cars, which would obviously take away from its ability to be used in other circumstances. However, I disagree with the title which states that the wheelcollider is entirely unusable; while it does have many problems, it is not to that extent. I have found that for many uses, it is in fact suitable, but the amount for which it is not is far more vast.
     
  4. pauloaguiar

    pauloaguiar

    Joined:
    May 13, 2009
    Posts:
    700
    The problem of the wheelCollider is setting and found the wheel current values for a simple 4 wheels car.
    like for example the car have the mass of 1200.
    Now it was necessary to find a formula of calculating the car's weight, and the wheelCollider settings for its corresponding value of the car.
    Example 2014 Toyota 4Runner: specifications all values is calculated for this type of Vehicle for example.
    http://www.edmunds.com/toyota/4runner/2014/features-specs/
     
  5. skirtz1

    skirtz1

    Joined:
    Mar 8, 2015
    Posts:
    20
    I agree with MT3 100%. Unity had never had a major component failing in such a big way. What is worrisome is the death silence from Unity on the subject. Please note that physics developers and some persons in charge stopped posting here! If they keep it quiet, to me that means there is no easy fix for this monumental blunder. Game development tool with poor physics engine - no kidding! PhysX 3 is not working, there is no going back to 2.8 (with some problems as well), and physics is broken (colliders, joints etc.) . Where do they go from here? I am not sure they know what to do at this point. Being quiet is their way to kick the can down the road... I am really worried it is downhill from here.

    PS:
    I would love for someone from Unity to come here and tell me that I am wrong, and what is the way forward. Please do!
     
    MT3 likes this.
  6. SnakeTheRipper

    SnakeTheRipper

    Joined:
    Dec 31, 2014
    Posts:
    136
    Has anybody got a good configuration to make it "work" realistically ?

    • How many torque on the wheels ?
    • The WheelCollider component values (slip extrem, assimptot, stiffness, etc...)
    • Mass of the car and of the wheels ?

    It would be nice to get some "standard" configuration and then slightly edit it to your own taste.
     
  7. terravires

    terravires

    Joined:
    Mar 27, 2013
    Posts:
    103
    Something does seem very off with WheelColliders. I've been working on a prototype of a tank and each time I try and tweak the settings things still don't "feel" correct. Either too much bounce and slip, or it just plain falls through the terrain. Then there are the times where it shoots off the map right after I hit play. The basic car example from sample project is a start. But if you try to add more axis/wheels and/or weight then things go crazy.
     
  8. Nition

    Nition

    Joined:
    Jul 4, 2012
    Posts:
    781
    No official bug fixes or changes to wheel colliders in Unity 5.1.
     
    MrEsquire likes this.
  9. RLin

    RLin

    Joined:
    May 10, 2015
    Posts:
    133
    I'm probably the only one on this forum who thinks this, but I believe that the wheelcollider is usable, with some very careful tweaking. Through all 7 pages, I haven't found a single person that liked anything at all about the wheelcolliders. The people on this forum sound like they're on the verge of forming a "Unity 5 Wheelcollider Hate Group."
     
  10. SnakeTheRipper

    SnakeTheRipper

    Joined:
    Dec 31, 2014
    Posts:
    136
    Show us those magic tweakings please.
     
    Tropicaya likes this.
  11. pauloaguiar

    pauloaguiar

    Joined:
    May 13, 2009
    Posts:
    700
    To prevent slipping and have better control play around whit the Stiffness for example front wheels: as forward stiffness : 1.8 and sideways stiffness 3.875. and for Back Wheels: as forward stiffness : 1.8 and sideways stiffness 4.875.
     
  12. RLin

    RLin

    Joined:
    May 10, 2015
    Posts:
    133
    That completely depends on the sort of vehicle you're making. It's pretty annoying, but you have to make completely new values for every different situation. For a Ford Taurus Police interceptor (or any sporty-ish sedan), I used 1500 mass, 20 mass per wheelcollider(4 total), 800-1100 torque(depending on gear), 4-wheel drive, and the default slip values with both forward and sideways friction stiffnes set to 1.5. All other wheelcollider settings are default.
     
  13. RLin

    RLin

    Joined:
    May 10, 2015
    Posts:
    133
    While I don't know what sort of vehicle you are making, those values seem very high for almost any land vehicle. If you use default slip values, I would stay at or below 2 for the stiffness values of both forward and sideways friction. I guess it is okay if you don't intend for your vehicle to behave realistically(maybe arcade-physics), but a real car would only surpass 2 if it was a non-production race car of some sort, or maybe a high-end super car.
     
    Last edited: Jun 10, 2015
  14. pauloaguiar

    pauloaguiar

    Joined:
    May 13, 2009
    Posts:
    700
    Yes i know the values depends of type of vehicle. That values i post is from my testing bad futurist car model, in this case works for this type, of course depending the mass in this car the mass is = 400 .
     

    Attached Files:

    • car.png
      car.png
      File size:
      1.3 MB
      Views:
      1,353
  15. RLin

    RLin

    Joined:
    May 10, 2015
    Posts:
    133
    What i meant is that such a high value would allow for little to no skidding at all, which isn't very realistic. You car would grip the road like it was glued at low speeds, and flip at anything slightly faster if you did not have a unrealistically lowered center of mass or artificial downforce.
     
  16. MrEsquire

    MrEsquire

    Joined:
    Nov 5, 2013
    Posts:
    2,712
    So what we are saying in conclusion, if you had a nice Vehicle Game and wished to upgrade from Unity 4.6x to Unity 5, only way your cars/vehicles are going to work as before is by manual tweaking for hours? or getting Asset author to add hack/do some kind of tweaking??
     
  17. RLin

    RLin

    Joined:
    May 10, 2015
    Posts:
    133
    It really depends on what you do with the wheelcolliders. Simple cars require very little tweaking, while things like tanks require more experimenting and maybe some special physics-altering scripts.
     
  18. JamesLeeNZ

    JamesLeeNZ

    Joined:
    Nov 15, 2011
    Posts:
    5,616
    so am I to understand the fix is in the suspension settings?

    tried upgrading my Inferno project (finally), and it mostly worked, except the wheel colliders.. cars just skittered around all over the place, even with no torque/brake applied.
     
  19. dylanjosh

    dylanjosh

    Joined:
    Jun 1, 2015
    Posts:
    10
    Hi, been trying for a week now to make wheel colliders work properly. Any help would be appreciated. I'm making a race bike.

    I'm having a lot of the same problems people are having on this thread. Sometimes after clicking Play, without any input the bikes starts slipping to the left or right. Tutorials aren't helping :(

    (I'm using 4 WC, just like car, but the gap between them much narrower)
     
  20. RLin

    RLin

    Joined:
    May 10, 2015
    Posts:
    133
    You have to make sure that the mass and suspension spring value are realistic relative to each other.
     
  21. JamesLeeNZ

    JamesLeeNZ

    Joined:
    Nov 15, 2011
    Posts:
    5,616
    I played around with them again last night and got them mostly working pretty well. Not as bad as I thought they were. Turned out my settings (from old colliders) for slip/spring/etc are now very wrong.

    Standard settings on a 1000 mass rigidbody worked fine.
     
  22. dylanjosh

    dylanjosh

    Joined:
    Jun 1, 2015
    Posts:
    10
    Bike is stable-ish now. But it's going rather slow, even with a forward torque of 50000. My Mass is set to 800.

    What could be causing this?

    The Wheel Colliders are all default values

    EDIT: Fixed by increasing stiffness in forward friction.

    But now my bike topples over on turning when it reaches high speeds. Changing center of mass doesn't help at all

    EDIT 2: Changing center of mass did help. Center of mass has to be entered manually in a vector 3. Weird stuff happens if you drag and drop.
     
    Last edited: Jun 18, 2015
    Dreamaster likes this.
  23. RLin

    RLin

    Joined:
    May 10, 2015
    Posts:
    133
    Of course your bike would tip if going fast, it's a bike. If it doesn't have a special balancing system it would behave like a kid riding a bike without training wheels for the first time.
     
  24. Roni92pl

    Roni92pl

    Joined:
    Jun 2, 2015
    Posts:
    396
    Hello, what kind of generic WC would you prefer to have - standard, Unity-like raycasted system, that apply forces to it's parent rigidbody and have suspension simulation, or have standard Unity's collider with shape of wheel, just with custom friction system. The second option would behave more properly in extreme situations, and you would have to create your suspension in car, like - wheel - hub - a-arm and connect a-arm with spring to car's body. What you think? I think, the second option is more realistic, easier to be done, and you could see whole car's suspension working(moving a-arm).
     
  25. Nition

    Nition

    Joined:
    Jul 4, 2012
    Posts:
    781
    A cylinder-shaped wheel collider would be ideal, but I can live with the raycast option.
     
  26. DrivingSim

    DrivingSim

    Joined:
    May 28, 2013
    Posts:
    32
    Here are my "magic numbers" which set all my AI Vehicle wheel colliders in the Start() function. All vehicles have a mass around 1350. This is what I came up with after weeks of fiddling. Read my post though. I have a specific use case. But maybe it'll be useful to someone or spark useful conversation from others who are lurking but not revealing their secrets. ;)

    Code (CSharp):
    1.     for (int i=0; i < _allWheelColliders.Count; i++)
    2.             {
    3.                 _allWheelColliders[i].mass = 10f;  // This can range up to 20 - effects braking a bit.
    4.                 _allWheelColliders[i].forceAppPointDistance = 0.1f;
    5.                 var spring = _allWheelColliders[i].suspensionSpring;
    6.                 spring.spring = 35000f;
    7.                 spring.damper = 7500f;
    8.                 spring.targetPosition = 0.5f;
    9.                 _allWheelColliders[i].suspensionSpring = spring;
    10.  
    11.                 var forwardWheelFriction = _allWheelColliders[i].forwardFriction;
    12.                 forwardWheelFriction.extremumSlip = 0.5f;
    13.                 forwardWheelFriction.extremumValue = 2f;
    14.                 forwardWheelFriction.asymptoteSlip = 0.8f;
    15.                 forwardWheelFriction.asymptoteValue = 0.5f;
    16.                 forwardWheelFriction.stiffness = 1;  // NOTE: 1-3 work depending on your need.
    17.                 _allWheelColliders[i].forwardFriction = forwardWheelFriction;
    18.  
    19.                 var sidewaysWheelFriction = _allWheelColliders[i].sidewaysFriction;
    20.                 sidewaysWheelFriction.extremumSlip = 0.5f;
    21.                 sidewaysWheelFriction.extremumValue = 1f;
    22.                 sidewaysWheelFriction.asymptoteSlip= 0.8f;
    23.                 sidewaysWheelFriction.asymptoteValue = 0.5f;
    24.                 sidewaysWheelFriction.stiffness = 5f; // If this is below like 3 or 4 vehicle tends to slide exspecially at higher speeds - like 55-60mph
    25.                 _allWheelColliders[i].sidewaysFriction = sidewaysWheelFriction;
    26.             }
    I've been watching this forum thread for awhile. At first I wasn't taking it too serious since it popped up so soon after U5 was released and I didn't feel the colliders were given a chance. A lot of the complaints I saw were the fact that all the settings had to be changed to get older projects to work again. It's a pain - I know - but sometimes that's what it takes to get improvements. However, after some serious ( and tedious ) testing I am convinced there's is some weirdness in the new wheelcolliders.

    I am working on a huge driving simulator about 8km x 8km world with about 50 AI vehicles that pool spawn around the player. We have lessons that teach driving. Some lessons let you drive freely, others are "rail-rides" as narration plays. So in the rail-rides the player is essential inside an AI vehicle - and this is where I noticed the weirdness.

    The AI cars follow normal traffic rules, stopping at lights, signs, etc. I need them to be precise so they don't blow through stop lights, etc. I had all the AI vehicle colliders dialed in for U4 and they worked great - nice and smooth.

    I had to change all their settings after upgrading to U5 and I was able to get them working again. When the wheels are set at what seems like normal U5 levels (as in my code settings) with stiffnesses = 1 the cars are too slippery especially around turns. They slide too much when braking, oversteer and fishtail. Only if I turn the stiffness way up to like 5 on both forward and sideways friction do they stick to the road. It makes no sense. You'd think the normal stiffness ranges would be between 0 - 1 or 2.

    However, when the stiffness is way up "rail-ride" lessons which have a first person camera are super jittery (This is not a camera issue!). So my next test was to lower the stiffness on the forwardFriction to 1 or 2 and leave the sideways high (5). This stopped some of the jitter but it's still there and noticable. This is as close a compromise as I can find so far. They don't fishtail and the jitter is minimized - but not acceptable.

    Also, I noticed that the wheelcolliders seem to shimmy even when the vehicle is at a full stop. I checked the Transform.positions on the root vehicle objects and it looks like a slot machine with values changing passed the third decimal point (or thousanth position) , so it looks as though there are calculations going on under the hood even when the vehicle is stopped. After extensive noodling, I can't seem to find the sweet spot with these new wheel colliders!

    We finally ended up sending our whole project to Unity support for help. I'll let you know if I learn anything.

    Our next step may be to puchase Edy's new U5 physics asset. We started with it on several projects back when U3 was out but found it too heavy on the CPU for our AI needs so we hacked and optimized a lighter version for the AI and left the original code on the player only which worked pretty good.

    On the latest project I spun my own code based off Unity's new car example because Edy was still working on his U5 update and I figured the Unity tutorial might reveal any new wheel settings and setup - which it did if you want to make a slippery racing game.
     
    Last edited: Jul 23, 2015
    reese01 likes this.
  27. JamesLeeNZ

    JamesLeeNZ

    Joined:
    Nov 15, 2011
    Posts:
    5,616
    Main thing I dont like about the wheel colliders is the numbers seem arbitrary...

    what does stiffness of 5 mean? I dont know... Its not stated in docs... its just a random magic number that is suitable.
     
  28. DrivingSim

    DrivingSim

    Joined:
    May 28, 2013
    Posts:
    32
    Agreed, I always assumed it was a multiplier on the settings above. so if your extremumSlip = 0.5 a stiffness of 5 would be 0.5 x 5 or 2.5f. But not a 100% sure there isn't something else going on under the hood.

    BTW: Even though our world is centered and the area the car drives is no more the 2000m out from center, I have wondered if some of my issues are related to floating point errors - not sure.
     
    Last edited: Jul 23, 2015
  29. LastLance

    LastLance

    Joined:
    Feb 22, 2015
    Posts:
    52
    Try ur settings, pathetic result... BTW, may be u know how to solve this - http://forum.unity3d.com/threads/wheelcollider-sleeps-at-work.343137/ ? Coz i'v spent 5~6h for fix, but, nothing works... Wheels "go sleep" in falling, but, suspension still works on grounded. Like a dubstep horse.
     
  30. DrivingSim

    DrivingSim

    Joined:
    May 28, 2013
    Posts:
    32
    Vehicles are really hard to diagnose on a forum without code examples and a specific setup explanation. Vehicle behavior depends on several factors such that any change in values of one can effect the others. If any of these are set up wrong it may not matter what wheelcollider settings you use - meaning the car will misbehave regardless if the wheelColliders are set correctly.

    • Vehicle Model Hierarchy
    • Ridgidbody mass
    • BoxCollider placement and size
    • Center of mass placement and making sure the rigidbody sees it...myRidgidbody.centerOfMass = centerOfMass.localPosition;
    • WheelCollider placement
    • WheelCollider mass
    • WheelCollider frictions settings

    Here is my vehicle hierarchy setup

    Chrysler200 (Ridgidbody & Boxcollider here- Box Collider size is about 2 x 1.3 x 4.6 (depending on car))
    AllWheelColliders
    WheelFL (WheelColliders here)
    WheelFR
    WheelRL
    WheelRR​

    AllWheelPivots
    PivotFL (These get rotated based on WheelColliders)
    WheelMeshFL​
    PivotFR
    WheelMeshFR​
    PivotRL
    WheelMeshRL​
    PivotRR
    WheelMeshRR​

    BodyMesh
    CenterOfMass (local position assigned to ridgidbody - generally at 0,0,0 for stability)


    General Component Settings for AI...on root car object

    Rigidbody
    • Mass: 1380 (for most cars)
    • Drag: 0.05
    • Angular Drag 0.1
    • Interpolate: Interpolate
    • Collision Detection: Continuous Dynamic

    Wheel Collider Settings on all wheels
    • Mass: 10 - 20
    • Radius: ~.35 (varies between cars)
    • Wheel Damp Rate: 0.25
    • Suspension Dist: 0.12
    • Force App Point Dist: 0.1
    • Center: 0,0,0
    Suspension Spring
    • Spring: 35000
    • Damper: 7500
    • Target: 0.5
    Forward Friction
    • Extremum Slip: 0.4
    • Extremum Val: 2
    • Asymptote Slip: 0.8
    • Asymptote Value: 0.5
    • Stiffness: 1 - 2
    Sideways Friction
    • Extremum Slip: 0.4
    • Extremum Val: 1
    • Asymptote Slip: 0.8
    • Asymptote Value: 0.5
    • Stiffness: 5
     
  31. RLin

    RLin

    Joined:
    May 10, 2015
    Posts:
    133
    Default settings actually work pretty well with stiffness set to 1.5 on both forward and sideways friction, rigidbody mass 1500, wheelcollider mass 100, and 20000 suspension spring with 2000 suspension damper. This functions relatively realistically.
     
  32. JamesLeeNZ

    JamesLeeNZ

    Joined:
    Nov 15, 2011
    Posts:
    5,616
    Settle down, I suspect PEBKAC.

    I can confirm the settings he used are fine, which mean the problem is your setup.
     
  33. RLin

    RLin

    Joined:
    May 10, 2015
    Posts:
    133
    @DrivingSim, if you want to use realistic values, you should at least look up your car. The chrysler 200 has a minimum mass of 1575 kg.
     
  34. RLin

    RLin

    Joined:
    May 10, 2015
    Posts:
    133
    A sideways stiffness of 5 is very high for those extremum and asymptote values if you want it to be realistic. Those values would probably give a very arcade-y driving feel.
     
  35. Roni92pl

    Roni92pl

    Joined:
    Jun 2, 2015
    Posts:
    396
    The biggest problem with WC is not how it reacts with some settings values, but instability and general wrong behaviour, especially in tough conditions(big mass, big friction values), and strange, very unintuitive implementation - it's based on sprung mass, which can be hard to understand for some people.
     
  36. DrivingSim

    DrivingSim

    Joined:
    May 28, 2013
    Posts:
    32
    Hmm. Might try the higher wheel mass - it effects the braking. Right now I have to set the brakeTorque around 600000 to get the cars to stop properly - I have a bunch of curves on the math though so they don't jam the brakes so may not work for everyone. However, with forwardfriction stiffness at 5 I only need a brakeTorque around 20000...kind of makes sense. I think the new torque uses actual Newtons.
     
  37. RLin

    RLin

    Joined:
    May 10, 2015
    Posts:
    133
    That still disagrees with the name of the thread, which states that the wheelcollider is completely unusable. While it has many problems, that doesn't mean its useless, unlike what most people seem to think.
     
  38. RLin

    RLin

    Joined:
    May 10, 2015
    Posts:
    133
    @DrivingSim, the key to wheelcolliders is not to use realistic values; it would be pretty insane for a sedan to be using 100 kg wheels. The values are pretty much completely irrelevant compared to their real-life applications. What you have to do is just play around with some (likely unrealistic) values until you get realistic behavior.
     
  39. DrivingSim

    DrivingSim

    Joined:
    May 28, 2013
    Posts:
    32
    That was my point in the original post...it was the only way I could get the vehicle to not fishtail - especially on the freeway at higher speeds. a stiffness of 5 made them look realistic on the freeway and worked well in the city/rural streets. I might be able to lower it a little but not much.
     
  40. JamesLeeNZ

    JamesLeeNZ

    Joined:
    Nov 15, 2011
    Posts:
    5,616
    What? Less than four posts ago you hassled him because his rigidbody mass wasnt the crysler standard...
     
  41. RLin

    RLin

    Joined:
    May 10, 2015
    Posts:
    133
    It fishtailed? How many degrees do your wheels turn, and do they use any smoothing in their rotation?
     
  42. RLin

    RLin

    Joined:
    May 10, 2015
    Posts:
    133
    @JamesLeeNZ, just to clarify, the vehicle rigidbody values have to be accurate, but almost none of the values on the wheelcolliders themselves line up with real life values.
     
  43. DrivingSim

    DrivingSim

    Joined:
    May 28, 2013
    Posts:
    32
    LOL, did that adnausum. Fiddled, faddled, noodled, and pulled my hair out. If you have magic numbers post them.
     
  44. RLin

    RLin

    Joined:
    May 10, 2015
    Posts:
    133
    I just did above.
     
  45. JamesLeeNZ

    JamesLeeNZ

    Joined:
    Nov 15, 2011
    Posts:
    5,616
    its a starting point, my standard mass is 1000 for my small vehicles. Complaining about mass then saying use made up numbers that feel real a few posts later is not helpful.
     
  46. RLin

    RLin

    Joined:
    May 10, 2015
    Posts:
    133
    Actually, it was more than four posts before you said this.
     
  47. DrivingSim

    DrivingSim

    Joined:
    May 28, 2013
    Posts:
    32
    I am using bezier path nodes to help point the wheels in the right direction using some vector math, the wheels don't turn much and are smoothed. If the smoothing is too low they oversteer the curve. If it's high they get a little jerky. It's a giant game of Jenga trying to balance everything.
     
  48. RLin

    RLin

    Joined:
    May 10, 2015
    Posts:
    133
    The main reason you want realistic mass values is so that the vehicles respond accurately to collisions with each other. I guess that you could scale everything down or up as long as they maintained the same mass relative to each other, but that would require some unnecessary math and adjustments.
     
  49. RLin

    RLin

    Joined:
    May 10, 2015
    Posts:
    133
    Really? I just use the super simple input axis, and changed the values so that the wheels would turn slower. I don't have any control issues that wouldn't exist in a real life car, but we should keep in mind that real life vehicles can flip over, skid and do other things relating to loss of control at surprisingly low speeds. You should also make the wheels have progressively lower turn speeds and maximum turn angles at higher speeds.
     
  50. DrivingSim

    DrivingSim

    Joined:
    May 28, 2013
    Posts:
    32
    Actually, every vehicle in my sim has the same mass with the exception of maybe large vehicles like buses or Mack trucks. I did this on purpose so that all 50 cars in the scene behave exactly the same while I tune the AI acceleration and braking curves and other behavior. I need a baseline to work with. We are not making this for NASA or anything so it's not that important to us. It only needs to look realistic when driving and I'm pretty close but wanted to see if there are any others out there with similar experience who can confirm my results. Also, I don't want to have to fine tune 30 unique vehicle settings because it causes timing issues and problems down the road, especially when doing "rail-ride" lessons. This is my third commercial driving sim and have learned where the pitfalls are for what we do. It's a compromise that works for us.