Avoiding rider ejections

[Vini]

Question for the bike modders:
Do you think that one could do something about the random rider ejections when the bike reaches very high lean angle and comes close to the ground (without losing traction or touching the ground, though)?

Maybe use some fake values somewhere to delay the moment of ejection?


It's quite ironic because a few people have been complaining about the lean angle lately and now we've got Manu's motogp bikeset and the bikes can reach much higher lean angle (just do some tests at suzuka and vallelunga) but the rider decides to just let go

[C21]

could you show/upload some examples?

[Vini]

sure

https://www.youtube.com/v/9XVtB4QoO9E
https://www.youtube.com/v/VwtZDvEBIYc


It is very easy to replicate at Vallelunga turn 4/5 with (all of) Manu's MotoGP bikes.

[Hawk]

Is that the knee slider that is hitting a bump in the track I wonder?

I know the knee slider appears to dump the rider when the knee slider hits ribbed kerbs on some tracks, so maybe it's doing the same when the knee slider hits a bump in the track surface?

Just some thoughts to ponder. 

Hawk.
PS: On your last video it looks like the collision mesh for the rider leg is possibly colliding with the collision mesh for the bike as the rider lifts his knee inwards.... Maybe this act is causing the rider dump?

[Yohji]

my mod also hppaened that.
I edit rider mass to little higher, maybe solve it.

[C21]

@vin97
in the first video the #46 Yamaha crash: could be caused by the incorrect calculation with banked corners...
#46 Murusama is a bit weird. Could be caused by incorrect track modelling or collosion with the banked corner marker from the rider....

second video: slowmo of the first one: banked corner , rider knee down, rider knee gets inwards again due to touch of the corner marker , rider "grip calculation"  thinks the lean angle is increased, sum of masses is exceeded , rider fall off.

pure assumptions of the videos 

[Vini]

I've done some more testing and I think Hawk is correct with this:

it looks like the collision mesh for the rider leg is possibly colliding with the collision mesh for the bike as the rider lifts his knee inwards

If you stop leaning the rider just before getting to the critical track area, this bug will not happen (I've even got the brake protector to scrap the curb without falling off).
If you are using auto rider lean, you will have to override the leaning direction to the opposite one to get the neutral rider position.
I don't know how often this bug happens to auto rider users because the automatic forward lean (depending on throttle input) might reduce the lateral leaning of the rider enough to avoid a crash at most corners.


Obviously, this rider ejection bug needs to be fixed in time for beta8.
Shouldn't be hard because you just have to disable bike<-->rider/knee clipping.


...I might post some more videos to prove this but you can also just test it yourself