I'm using Bullet 2.73 with a genetic algorithm to evolve movement patterns for a model of a snake robot. The main problem I'm having at the moment is that for the same movement pattern I get different results every time. I read on a post that it was possible to disable solver randomisation to get deterministic results. How do I go about doing this. Also what is the best way to reset the simulation.
Thanks for any help,
Adam
Deterministic Simulation
Re: Deterministic Simulation
In my tests with Bullet I got always the same result... look at the following link about Determinism and check if your gamecode does the same:
http://www.bulletphysics.com/mediawiki- ... eterminism
Cheers,
Gilberto
http://www.bulletphysics.com/mediawiki- ... eterminism
Cheers,
Gilberto
Re: Deterministic Simulation
Im using a fixed step size I think that might be the problem. I don't really have a choice because I need to be able to run the simulation as fast as possible with the graphics turned off for the genetic algorithm and then run the simulation with graphics turned on to view the results. My model is fairly complicated with 49 rigid bodies and 64 constraints, I get similar results with the same movement instructions but not exactly. Thanks for the help anyway.