How towr calculate the force？

[littleggghost]

Hi, @awinkler
Last time we talked about the dynamic simulation.
I notice that your achievement in latest paper is like the video by Emo Todorov, University of Washington. The goal is also set up the initial condition and final condition and the optimizer will solve the motion planning, but the difference is that he used a kind of reinforcement learning algorithm. He used a physics engine called MuJoCo physics engine.
The well-known physics engine is like Bullet, ODE, Vortex, etc. I read the towr_core and found lots of functions for dynamics. Could you tell me how to calculate the force except using the physics engine in towr?

[awinkler]

Hi! The dynamics (+ forces ) are enforced through constraints in the optimization problem. The keyword here is Trajectory Optimization, with shooting and direct methods. Here's an overview.