Strategies for simulating pedestrian navigation with multiple reinforcement learning agents