A Comprehensive Study on Drone-Based Monitoring of Palm Fruits Ripeness for Sustainable Harvesting

Accurate and timely assessment of palm fruit ripeness is critical for maximizing oil yield, reducing post-harvest losses, and promoting sustainable harvesting practices. Traditional ripeness evaluation methods rely heavily on manual inspection, which is labor-intensive, time-consuming, and prone to human error. This study presents a novel drone-based monitoring framework that integrates LiDAR (Light Detection and Ranging) 3D mapping with deep learning-based object detection—specifically the You Only Look Once (YOLO) convolutional neural network architecture—to automate and enhance palm fruit ripeness classification. Unmanned aerial vehicles (UAVs) equipped with LiDAR sensors and high-resolution RGB cameras were deployed over oil palm plantations to acquire dense 3D point cloud data and aerial imagery. The collected data were processed through a YOLO-based detection pipeline trained on a curated dataset of 4,200 annotated palm fruit bunch images spanning three ripeness categories: unripe, ripe, and overripe. LiDAR-derived structural features including bunch height, canopy density, and spatial distribution were fused with image-based color and texture features to improve classification accuracy.