Urban autonomous driving decision making is challenging due to complex road
geometry and multi-agent interactions. Current decision making methods are
mostly manually designing the driving policy, which might result in sub-optimal
solutions and is expensive to develop, generalize and m