ALGAMES: a fast augmented Lagrangian solver for constrained dynamic games

Dynamic games are an effective paradigm for dealing with the control of multiple interacting actors. This paper introduces augmented Lagrangian GAME-theoretic solver (ALGAMES), a that handles trajectory-optimization problems actors and general nonlinear state input constraints. Its novelty resides in satisfying first-order optimality conditions quasi-Newton root-finding algorithm rigorously enforcing constraints using method. We evaluate our context autonomous driving on scenarios strong level interactions between vehicles. assess robustness Monte Carlo simulations. It is able to reliably solve complex like ramp merging three vehicles times faster than state-of-the-art DDP-based approach. A model-predictive (MPC) implementation algorithm, running at more 60 Hz, demonstrates ALGAMES’ ability mitigate “frozen robot” problem onto crowded highway.