Beach environments face a growing threat from accumulating solid waste, particularly plastic debris. This waste originates from improper disposal practices and travels through rivers and streams to accumulate on shorelines. Traditional collection methods, often reliant on human intervention, pose challenges in such dynamic environments. This paper proposes an autonomous waste detection and collection system specifically designed for navigating sandy beaches. The system leverages a MobileNet SSD-based object detection and recognition model for real-time waste identification within the rover's path. A robust distance estimation algorithm is implemented to ensure efficient collection and safe navigation. This algorithm plays a critical role by providing real-time data on the proximity of detected waste objects, enabling the system to optimise its pick-up manoeuvres. Additionally, an obstacle detection and pick-up mechanism is incorporated to ensure safe and efficient waste collection.