Abstract

ProSeCo Planning is a probabilistic, semantic, and cooperative planning library that enables implicit cooperative decision-making for automated vehicles in heterogeneous environments. At its core, a Monte Carlo Tree Search (MCTS) is used to find optimal trajectories in challenging multi-agent driving scenarios. The base MCTS is improved through various extensions, generating stronger solutions in a shorter time.