VERLET INTEGRATION - couple of questions

Blitz3D Forums/Blitz3D Programming/VERLET INTEGRATION - couple of questions

Stevie G(Posted 2005) [#1]
I think I've had a look at almost every verlet system available in the code archives and decided to have a go myself.
The basics are relatively straighforward and I've managed to successfully create a rigid body tank
moving over a custom terrain using blitz collisions applied to each particle.

1---2
!\!/!
!-0-!
!/!\!
4---3


Using a dual analogue control system ( one for each track ) it was quite simple to apply forces to the particles,
representing the tracks which had collided with the terrain (1&4)+(2&3) to allow for both rotation and forward / backward movement.

What I really want to be able to do is create more of a rear wheel drive car. I know I can apply a forward force to
the rear wheels for straight movement - I'm kind of stumped with the steering?!

Assuming the body is moving upwards, where particle 0 is the centre of the body, particle 1&2
represent the front wheels, 3&4 represent the rear wheels and particles 5-8 form the top of the body.

Note that the front wheels can turn @ +-45 degrees, local to the body - how do I simulate the steering / assuming
the force is only coming from the rear wheels? Assume I have to distribute the force applied to the rear wheels on to
the front somehow? I also need to be able to turn the car realistically when no acceleration is being applied and avoid
hovercraft style movement - i.e. apply greater friction to the particles not moving in the forward direction?

I'm not asking for anyone to write this for me - I'm really looking for some pointers from those who have done
something similar and overcome the intial headaches - so any help is much appreciated.



Cheers
Stevie


Techlord(Posted 2005) [#2]
Stevie G

I gave Verlet Integration a go a while back. It merged BVH Boneset, Verlets, and Springs into one system. However, I placed it back on the burner.

It would be nice to see a high performance physics system made with Blitz3D.


Clarks(Posted 2005) [#3]
verlets are good but rotations are very difficult to calculate.


Stevie G(Posted 2005) [#4]
Thanks folks - this doesn't really tell me what I need to know though :(


flying willy(Posted 2005) [#5]
Another astounding post by clarks.


Techlord(Posted 2005) [#6]
Stevie G

Sorry, I failed to read the entire post. I really didn't get as far as you have. I'm not sure that anyone has with a blitz-made verlet system.

Considering a tank is a all wheel traction vehicle, you may to have to contend with both front and back steer. My approach to a tank would have been much different only using the verlet for the chassis, springs for the L/R Front & L/R Rear suspension. I would totally fake movement and steering (rotation) from the center of the body with the standard translation functions. Stiff movement and steering with a tank is ok.


Stevie G(Posted 2005) [#7]
Thanks Frank,

No problems with the tank - I only apply force to the track which is moving and touching the ground, therefore if you only push the right analogue forward it turns slowly to the left, if you move the left analogue back and the right forward it spins on the spot to the left. It works quite well so this isn't the issue. I would however be interested on how you would bring springs into the equation - assume for some form of suspension? Would they work in a similar fashion to verlets but with an elasticity factor for the contraint?

The question is really about simulating a car rather than a tank but I'm interested in all aspects.