Reinforcement Learning-Based Autonomous Navigation and Obstacle Avoidance for USVs under Partially Observable Conditions

Figure 4

The overall GRU processing flow, whose generated contains previous state information that improves the USV’s decision accuracy.