Obstacle avoidance navigation for an unmanned surface vessel is a research focus for ship autonomy in which the real-time requirement in practical application is very serious, and always necessitates a complicated structure model to guarantee real-time performance. This paper proposes the grid cell activation model to reduce the complexity of modelling and to simplify an obstacle avoidance algorithm. Combined with the goal-oriented probability model to design a dynamic positive-loss-rate expectation evaluation function, it produces the proper strategy for obstacle avoidance. Case studies on multi-obstacle layouts and special circumstances are conducted and presented. The results indicate that the grid cell obstacle avoidance algorithm can effectively implement obstacle avoidance planning and ensure real-time requirements. A comparison with the potential field algorithm is performed, which shows good results and verifies the feasibility of the algorithm.

中文翻译：

---

基于网格单元机制的无人水面舰艇实时避障规划研究

无人水面舰艇避障导航是船舶自主的研究热点，在实际应用中对实时性要求非常高，总是需要复杂的结构模型来保证实时性。本文提出了网格单元激活模型，以降低建模的复杂度并简化避障算法。结合面向目标的概率模型，设计了一个动态的正损失率期望评估函数，产生了合适的避障策略。对多障碍布局和特殊情况进行了案例研究并进行了介绍。结果表明，网格单元避障算法可以有效地实现避障规划，保证实时性要求。