Joints stiffness

[JoeJ]

hehe, seems quite complicated and i don't have the necessary the vocabulary
Just tried again and the truck can always escape now.

I'm thinking of a making race game along engine and editor developement starting in few months.
Probably just for fun, but who knows...
One idea i have is to use a ramp to get up on 2 wheels on the side so you can drive a narrow shortcut path for example.
Also i want very high speed (1000 km/h), accelerator tracks, loopings and that kind of fun stuff.
Might become challenging to do this ontop a accurate simulation...

[Sweenie]

One idea i have is to use a ramp to get up on 2 wheels on the side so you can drive a narrow shortcut path for example.

Not at 1000 km/h I hope
Actually, that would be fun to try.

[Julio Jerez]

Ok Joe and Sweenie I now added the solve that I needed to do the Balancing rag doll.
This allow for a lot of features that can be implemented on the Tool side and the application side of the engine.
To test it I now complete the positronic slip differential for the vehicle that Ms Vito explain on the movie My Cousin Vinny.
I was looking for a video that explain exactly hwo is support to work and I found one
https://www.youtube.com/watch?v=4U1SLSV9wNk
https://www.youtube.com/watch?v=RcRBYpccP6g

Oh and I forget, like the guy in the video say, we also have support for independent axel differential wheel lock, but for now I only expose global control, which I believe is enough, But if anyone make a rock climb game, the option is available.

Basically now our vehicle will run with normal differential, but on demand the play can call the function Lock Deferential and the speed difference is reduces to not more that 5 rps.
These mean that differential is always active but is has too setting.

if you guy test it, please see if you can reproduce the effect in the video.

the key to lock the differential is 'L'
It also show a different handling with the option on and off, In normal drive the handling is better when the differential is unlocked.

I suspected that a differential was a big part of the driving behavior, but never I thought it was that much.

Anyway this allowed me to stress test the solve and now I am fully ready for the Balancing ragdoll

[JoeJ]

So this limits the maximum difference of all wheels rotation - that's what i've had in mind to fix a 'got stuck because one wheel is in air' problem.
Works good
But because i know nothing about cars and related terminology yet i can't give better feedback than that.

I remaining problem to me is that wheels rotate pretty fast most times and slip over the ground.
I'd expect better sync here between wheel and ground - more kind of targeting static friction.
Probably just a matter of settings, but that's what a noob like me sees.

[Julio Jerez]

I remaining problem to me is that wheels rotate pretty fast most times and slip over the ground.
I'd expect better sync here between wheel and ground - more kind of targeting static friction.
...

actually that is the intention, the engine in eh vehicle is actually very powerful, because ether are super car. I am driving a full throttle intentionally so that the tire get saturated and spin so that the can car lose traction and skip and drift.
It is not is no an overlook is intended. check outer any video of a Monster truck or completion of car drifting.

is actually quite easy to drive with tire locked to the ground, one placing a less powerful engine, maybe half the power, or having a analogous steering which I do no have the ability wit the version of wxWidge.
The power is what make super cart difficult to drive. because the tire skid a lot under any pedal input.

[JoeJ]

Ok, as expected.
I'll try to make automaic throttle control for both easy driving and powerful engine, i don't believe in analog joystick control for throttle is a good thing for arcade.
But i want to use analog steering by mouse - no game got that working except Indy 500 from the 90's.
That was so much better than a joysteck or wheel that pulls you back to zero all the time.

[Sweenie]

Excellent!

Can confirm the differential lock makes a difference.



It actually works better than the video shows, I messed the recording up by getting stuck on the blue barrier and releasing the gas. But the vehicle still had no problem climbing over without having to get a running start as the first vehicle have to.

[Julio Jerez]

Nice,
A Ray cast car will never do that. with this model we may have to work harder to get it on, but is worthy because you can read about vehicle physics and, using abstraction is possible to get a plausible simulation. Ray cats cat is just craps. I am glad I decided this rout.

The next step is the lateral stability for been able to steer the car driving sideway.
The code for that is part in but I has not tested yet because this has the ability to hide bugs.

the reason Is that lateral stability is all about small perturbation of a steady state.
Basically what this means is that giving a state of a set of variables.
What changes can be applied to all the variables until the state loses equilibrium.
the equilibrium is measure by the total potential energy in the system.
if the perturbation of the set of variables increase the potential energy the object is steady equilibrium. that is when a car drives with a slip angle that move in the same direction that the tire slip angle. That almost every vehicle.

The interesting part is when you have a state if unstable equilibrium here a small change decreases the potential energy. and the can lose control. but if you do just right then the car move side way and doo all the cool drifting.
This is hard to do, an dis what I am going next.

[Julio Jerez]

Ok, as expected.
I'll try to make automaic throttle control for both easy driving and powerful engine, i don't believe in analog joystick control for throttle is a good thing for arcade.

I see what you mean, is not fun if you get all the power at once. but the good new is that now that we have a versatile solve with can plug anything we want.

Until now I was trying to emulate engine friction by using viscus damping in replacement of dry friction.
Basically when we nee is a Gas valve regulator that passed gass proportional to some settings.
This is hard to simulate with velocity base friction when there are many part interconnected, but is easy with dry friction.
Basically we add a joint between the engine and the chassis, that limit the shaft acceleration to a factor of the gas valve. This will prevent the rush of power when the key is pressed. It will work nice analogues input, plus can be configures by the player. Hey we can even emulate realistic Turbo.

Another thing I will also add is the internal gear friction, for the longer time it always bothered me that each time I made a vehicle model, the vehicle lost a lot of speed each time you turn. Must people say this is Akerman angles problem, but I never really believed that .
I always suspected that this is due to tires wanting to keep angular momentum, and that only a kinematic differential can solve that. I think I was right however the results better than I was expecting. Basically the car now can coast and barely loses any speed when turning, this is not good either.
I realize that cars do lose speed but is because the internal gear friction more than anything. Now we can simulate that too.

so I will try to add that tonight.

[Julio Jerez]

Hey Joe, I found the bug that made the motor jitter.
Basically when I found a round that violate the friction forces, I was moving tat row to the inactive variable set, by is doing so, I only keep the force value and no the acceleration.

In general the acceleration is rarely use, by f you have several variable mining in and out for the inactive set, the acceleration has to be update as well, that was why it was jetting

in one pass if found one variable and if on eh next it turn out that was no the optimal variable is has to retrieve the one from the native set and plus the new one, however the old one will have the wrong acceleration.

The was the reason for the jitter, now I will complete the motors stuff so that will would never have to worry about that.

[JoeJ]

Wow, this looks really good now
Big difference - just natural.

I'll be busy for some days, then i'll get into my ragdoll again and see if it works here too now...

[misho]

Is this relevant to my problem of rigid joints being unstable at high timestep values?

Misho