Learning to Drive in the NGSIM Simulator Using Proximal Policy Optimization

Figure 2

The architecture of the actor-critic.