Enhancing Beach Cleaning Efficiency: ML-Based Litter Detection by an Autonomous Robot

Beaches are an essential interface between the land and water and act as a shelter toward the ecosystem which comprises various plants and animals. With the growing rate of tourism across the globe, beaches continue to grow in popularity among the population as a recreational area to visit most times during the year. Which has led to the rapid increase in litter on beaches that severely affects the coastal communities for their livelihood. Municipalities have begun taking various initiatives toward reducing litter found on beaches. To tackle the litter found on beaches, it is imperative to address the underlying root causes through the implementation of efficient and innovative methods of beach cleaning. This paper presents litter detection technology to be used in autonomous robots deployed in beach cleaning operations. The system is based on a machine learning-based approach using YOLOv8, that detects and counts the litter observed by the autonomous robot. The object detection model is trained on separate categories of commonly found litter. Once an object of interest passes through a predefined section of the frame, it updates the counter. The obtained data can be analyzed and presented to the user on a dashboard that can be accessed from anywhere around the world.