Real-Time Physics Simulation Forum

It's deactivation. After some time being inactive body goes to "sleep". Use body->activate(true) before applying any forces to make sure it's not sleeping.

This would lead to jittering and instability, you wouldn't be able to find correct coefficients for motors, and there will be an accumulating offset in every frame. This is how I actually got to my solution — motors are just a way to set velocities, so why not set it directly.

I have tried this approach to create wheel brakes, but it doesn't work. I ended up calculating the amount of distance a body moved from it's equilibrium and setting it's velocity to (-distance / fps).

For some reason btGeneric6DofSpringConstraint does have only static (velocity-independent) friction. I"d suggest to subclass btGeneric6DofSpringConstraint and modify friction force calculation. There is even an example on this forum somewhere.

It works well in FPSs' without simulated rigid bodies. With the "real" physics, however, even 100ms lag would make user notice how his boxes are penetrating a floor. With some big random network lags the game would be totally unplayable. When using a local physics simulation even occasiona...

Let's see. In an ideal case there will be a 100ms lag at best. It is a very noticeable delay, so you will probably want to implement some extrapolation which can be good in only one case: using the same physical engine as server does. So it would be in the end the same technique the most multiplayer...

A strange thing happens, when I change the fixedTimeStep (3rd parameter of stepSimulation) to some small value (1 / 400, for example). All spring constraints seem start to have more "strength". I adjust erp parameter: float physicsResolution = 1.0f / 60.0f / 10.0f; float dt = timeSinceLast...

PhysX does it too, but probably faster, since it runs very smooth on mobiles.

I've managed to emulate static friction by remembering constraint angle when wheel's angular velocity reaches small threshold or changes direction, then setting it's angular velocity to [(angle - angle_static) / fps]. It's not perfect, the wheel still has some difference from the state it should be ...

There is a custom vehicle made of rigid bodies and joints, wheels are cylinder shapes connected to body using hinge constraints. I can't figure out how to implement brakes. Wheels have very large friction coefficient, and the vehicle must be able to stand still on very steep (almost vertical) surfac...

It seems that btGenerateInternalEdgeInfo locks vertex buffer of striding mesh interface twice: Before the (triangleIndex, numFaces) loop: ... meshInterface->getLockedReadOnlyVertexIndexBase(&vertexbase,numverts, type,stride,&indexbase,indexstride,numfaces,indicestype,partId); ... then, in th...

The moving object is a constraint-based vehicle, it has springs to damp bumps on ground's surface. The ground is represented by static triangle mesh shape (bvh). I use btInternalEdge utility, but wheels still do bounce on edges of perfectly flat triangle surface. Maybe enabling split impulse helps a...

Thank you, it gave me a direction to think on.

And how do you handle rolling objects? They may have opposite velocities, but every contact point will be fixed in objects' local space (if there is no slipping).

These are constraint points in objects' local space.

I've managed to do simple contact sound effects (iterating over contact points every simulation tick and taking m_appliedImulse). But how can I implement friction sound (e.g. when vehicle tires slide or a crate being dragged)?