Articulated Body Dynamics Simulation using Optimization Integrator

We present a novel optimization-based algorithm for articulated body dynamics simulation. We formulate the governing equations of rigid body dynamics using only the position variables and recast the position-based articulated dynamics as an optimization problem. We also extend our framework to handle joint limits, control forces/torques, fluid drag forces, and frictional contact forces. Our reformulation allows us to use an off-the-shelf optimization algorithm to time integrate the articulated body with an arbitrarily large timestep size and analyze the stability. Our algorithm can efficiently perform each iteration of optimization within $\mathcal{O}(N)$ time using Quasi-Newton method and $\mathcal{O}(N^2)$ time using Newton's method, where $N$ is the number of links. We highlight the performance on different benchmarks and compare the performance with prior articulated body dynamics simulators.