Tire material from newton 1.xx, multibody vehicle

[PJani]

I am loosing my mind with tire friction and multibody vehicle. It seems that, i am getting no where, my vehicle is not behaving as it should, anything i do turn out to be a disaster... :/

Julio is possible to get a peek at MultiBody Vehicle callback for tire material from newton 1.xx? Just callback part nothing else or at least some tip how did you make callback for tire material? Because i cant figure out why the wheels in newton 1.xx where so smooth.

[PJani]

Anyway Julio did u use any special materialpair callback for tires in newton 1.xx?

I have newton 2.24. And my wheels are skiding like hell. My materialpair has settings kinetic 0.4 and static 0.9.

[PJani]

One thing i only want to know how were tires( in this function NewtonVehicleAddTire) implemented in newton 1.xx? Convex cast, Raycast, Body?

[Julio Jerez]

The Mutibody joint in 2.0 is the same joint that was use in 1.53.
There is only one difference. in 1.53 the engine simulate the car at a higher fps.
this is similar to what the engine does now when using continues collision. and a body move a high speed.
I supersample the movemnet to find the closet contact point, at a relativally decent swep collision..
with the car, it calculates the simulation frequency so that the tires moves at more of less constant speed over the floor.

This give more friction sample points when the car moves at highet speed, and better grip response.
You can see that behavior in the old 1.53 by rendering the contacts points at the tires.
you will see they there are 2, 3, or 4 contacts depending how fast the tire is rolling.
It is that supersampling function that gives the grip feeling to the car in 1.53.
In 2.0 the funtionality is not exposed. but I can expose it for some features since it is there.

I did not exposed at the time because when I was working in the multibody car, i decided to switch to the Raycast car and now I have neither.
Like I said before, I like to continue this but, in the last three month I started the editor, and on top of that I am in a crunch time at work,
so I can not make too much progress during the weeks day, only on weekends.

All in All, what I am saying is that yes I will complete both car joints, but that will be after I get the File format and the editor operational,
Because it is too wastful to work procedurally on a features and then have about 2/3 of the end users not using it at all because it is too hard to undertand.
Even I get confused with the code after I do not use it for a long time.

[PJani]

Thnx i am looking forward to see that in action. Keep a good work.

[Sash]

Yes, I'm waiting for either MB/RC vehicle joints too....
For now I have managed to get it working close to "acceptable".
My solution - to use helper forces for acceleration, braking and sidesleep compensation.

http://www.youtube.com/watch?v=MZ8I6Xg54HQ

I still have only 2 unresolved problems.
- unstable suspension (especially at zero speed)
- strange AddTire behavoir (seems like they added with global (although claimed to be local) coordinates) when I'm using Z Vector for front - so GetSpeed() doesnt work too (in fact not a big problem).

We're all counting on you, Julio

[Julio Jerez]

hey Sash that is not bad. I like the obstacle track and when the car hit the curve and flipped over.

I am having the hope that all thso eissue will be resulve when I am ready to make car by blu prints rather than By code.
I beleive that in tow or three weekends from now we will be ready to able to do that.

Actually I has some great tips for Rvangel the man who make Racer, that I will apply to the car next time I am working on it.

[JernejL]

Wow i wouldn't want to drive that car, the suspension is almost totally rigid

[Sash]

Wow i wouldn't want to drive that car, the suspension is almost totally rigid

Yes, I dont' like it too. In fact this is juat a lack of spring strength. Otherwise, when increasing springConst suspension will explode.
Effect is especially noticeblae @ low fps. So I switched to VSync 60fps and lowered springConst to avoid it, while feeling suspension still works.
Also, I noticed suspension is more stable, when redestributing masses from RL values to "heavier wheels, lighter body" setup.

On other hand I put this demo mostly to demonstrate how do I fight with low tires' grip.
When using helper forces my vehicle is capable to run 120-150 Kmh and still responds on steering realistically, while with "just wheel torque" approach she is uncontrollable (both speed and steering) at 30-40 kmh.

[PJani]

I found out that using SM_EXACT(0) solver the whole vehicle starts to be unstable so i am using SM_FASTEST(1). Temporary i was using high friction values over 1.0(16.0 for static and 14.0 for kinetic). I used that with that Ural(). For acceleration and deacceleration and doing turns in slow speed there were no problem. When i was doing turns in high speed some weird and annoying effects where produced(truck started to jump, like kangaroo).