# I The problem of simulating absolute grip

Tags:
1. Apr 10, 2017

### quickquestion

Ok, so I'm not even bothering with tire friction simulation anymore. What I'm trying to simulate is a tire that has 100% grip at all times (no lateral velocity.) Now you see that the statement I just made has a contradiction in it...If a car has 4 tires, and if a car is rotating, it is impossible for all 4 tires to have zero lateral velocity (unless all 4 tires have ackerman steering.)
However, using ackerman steering would simply be a cop-out way for never really having to solve the underlying problems.
So basically, I am going to list the problems with simulating absolute grip here (For now we are completely ignoring friction equations and just trying to have tires that try their best to grip 100%.)

to explain problem 1 more in depth, the problem is phy_speed updates only after the physics has been applied, so that the entity must first move forward before it can change the lateral velocity of the tires.

To explain more of problem 2 more, the overall goal is to cancel all lateral velocity of the tire...This results in pure longitudal velocity for each tire...BUT if this is the goal...then look at the paralelogram location of the tires...the actual destination of the tires cannot be this, because each tire follows a curve with a radius, the path of the inside tire cannot be same length as the outside tire...thus simply putting the tires where the pure longitudal velocity indicates, cannot possibly be the right method.

I am really at a loss. Any Brian Beckman's in the audience?

Last edited: Apr 10, 2017
2. Apr 10, 2017

### Nidum

Wheel speed difference . Inside pair v outside pair .

3. Apr 10, 2017

### A.T.

Have you actually tried doing a web search on car physics simulation? There seems to be plenty out there, including example code etc.

4. Apr 10, 2017

### quickquestion

Yes, but we are dealing with Box 2D physics, it has no such thing as wheels. All we have to work with is abstract thrust vectors. So in terms of Box 2D, we are looking to cancel the sideways speed of the wheels, and I'm not sure how knowing the inside longitudal speed will give us the needed sideways force. Basically I will say exactly what I am after: Basically I am trying to find an equation that will tell me the correct amount of sideways force to apply to each wheel. So that right there basically sums up what I'm after.

Have done tons of searching, none really suited my fancy. Some examples were better than others, Iforce 2D was pretty bad and basic, others were more advanced but still not quite right or realistic. I stumbled upon Brian Beckman's car physics and he said it was basically not an exact science, and he was the first guy to discover the "combined grip/slip" equation. He said he wasn't sure his math was right (this document was 10 years ago.) The reason for this is I suspect most fall prey to the Newtonian physics friction problems...newtonian models give an incorrect analysis of friction, In real life a car-wheel (longitudally), behaves like a cog on a toothed rail. But laterally, it behaves like a damping field. Newtonian model attempts to crudely represent behaviors as a thrust force (which, unfortunately, thrust forces are the pretty much the only available functions available to my physics engine.)

That being said, what particular car physics examples do you recommend for me?