To address the issues of slow convergence speed and high path costs in the RRT* algorithm. This paper proposes a Quick Potential-Biased Hybrid Sampling RRT* (Quick-PBHS-RRT) algorithm. Building upon the RRT* framework, the proposed method integrates three sampling strategies: uniform sampling, Gaussian sampling, and goal-biased sampling. It further incorporates an artificial potential field (APF) for trajectory guidance while implementing a deep node optimization mechanism. Comparative simulations in identical environments demonstrate significant improvements: Compared with APF-RRT* and Quick-RRT* (Q-RRT*), Quick-PBHS-RRT* achieves 11.8% higher path quality than APF-RRT* and 0.69% improvement over Q-RRT*. Regarding computational efficiency, the algorithm reduces convergence time by 89.3% relative to APF-RRT* and 76.9% compared to Q-RRT*. Finally, the robustness of the algorithm is further validated through line-of-sight (LOS) navigation experiments implemented on a manta robotic vehicle platform.