<p>Unmanned Aerial Vehicles (UAVs) are increasingly being deployed for applications such as surveillance, search and rescue, secret operations, traffic monitoring etc. However, autonomous path planning in dynamic and unknown environments remains a challenge. This paper presents a novel approach for UAV path planning using ultrasonic sensor data and the Soft Actor-Critic (SAC) reinforcement learning algorithm Enhanced with a Prioritized Replay Buffer (EPRB). Unlike standard SAC, which samples past experiences uniformly, the proposed method prioritizes transitions with higher learning potential; typically those with larger Temporal-Difference (TD) errors. This allows the agent to focus on more informative experiences, accelerating convergence and improving policy quality. As a result, the improved SAC-EPRB framework demonstrates faster learning, better sample efficiency, and improved obstacle avoidance and goal-reaching performance in dynamic environments. The implementation is carried out in MATLAB Simulink, providing a realistic simulation environment for UAV dynamics and sensor integration. For performance evaluation a comparative analysis of the proposed method is carried out with the state-of-the-art methods. Experimental results demonstrate that the SAC-EPRB approach outperforms traditional reinforcement learning methods in terms of path optimality, collision avoidance, and computational efficiency.</p>