I know this is an older thread, but appears people still refer to it, so in the spirit of transparency – per-point collision impulses are available since 2022.1.0a10. @Edy
Thank you, post updated 
Has PxVehicleSuspensionSweeps been implemented yet?
Nope.
@yant any update on PxVehicleSuspensionSweeps? Anyone working on it?
Does this mean this is now available by default in Vehicle Physics Pro?
Because Im using it and I still see the frontal collisions for the wheel collider are noy realistic. Seems its still using just one point for ground detection from the center of the wheel.
In Vehicle Physics Pro I’ve abandoned the Unity WheelCollider completely and replaced it with my own wheel contact model that just works as expected.
This I think is somewhat related and hence I think the best place to post this under, WheelCollider.suspensionExpansionLimited when enabled, seems to have a bunch of weird behaviours.
What appears to be a bug is that if I turn one of my cars basically at all(any amount of speed), the front inside wheel’s wheelcollider.rpm will report a flat 0, despite being able to see the gizmo of the wheel in question rotating. The reported slip value is also extremely low, though the groundhit does return, implying the wheel thinks it is grounded, so it’s not being set fully airborne(if it were to get airborne and still be rotating, ie off a jump, the rpm value doesn’t go to 0 anyhow).
That’s a 1600 kg vehicle with very soft suspension. On a 1400 kg car with slightly stiffer suspension, the problem persists but only at lower speed, where the outside suspension would be less compressed. In any of the listed cases, there’s more than enough suspension travel to put that wheel on the ground. Stiffer/lighter cars don’t suffer from this at all, where you’d actually expect the stiffer spring to be more resistant to expansion.
I’m puzzled as to why this would happen, given the docs state the boolean would affect “extreme values for Suspension Spring properties”, and catching air. In none of the cases do the wheels report being off the ground, and all of them are softer sprung vehicles with spring values around(or below) the defaults, unless ‘extreme’ means soft for the vehicle’s total weight, and a rotating wheel reporting 0 rpm is a bug regardless of any other part of that.
Secondarily, braking is a lot worse, on all cars, from the 1600 kg one with soft suspension to a 900 kg lemans racer with stiffer suspension. I would expect, if anything, for this boolean to make the rear of the car lift more, and load up the front tires extra. Considering all of the cars can hit ABS(applied per wheel) with or without the bool, more load on the fronts should trivially increase steering authority under brakes as well as probably the amount of braking the car can do in total, but does neither, which is puzzling.
I’d love to run this variable as I want to have a suspension that moves about for gamefeel and realism reasons, and I can see it makes the car likelier to tip when hitting a curb at high speed, which is what I expected to see, but the above listed unexpected behaviours and lack of any proper documentation make it a non-starter.
Further, I enabled this specifically because I noticed that if I start doing a burnout from a stop and roll forward while turning(front wheels locked via brakes), the car will veer to the opposite direction to where it is turned, but that seems to be a wheel slip model issue, this bool does not fix it.