Fostering Precision Agriculture through Affordable 3D LiDAR Deskewing with EKF

Abstract

Precise 3D point cloud mapping plays a crucial role in a variety of agricultural robotics tasks such as field planning, navigation, harvesting, and monitoring crops. However, providing a sensibly accurate representation of the surroundings using a robust, cost-effective solution that minimizes computational and financial burdens poses a significant challenge. This study presents a novel solution to acquire 3D point cloud data using a low-cost 3D LiDAR scanner, relying on a model to de-skew the sequences of 3D point cloud representation using non-holonomic robot motion. In order to correct point cloud distortions effectively, we propose an Extended Kalman Filter (EKF) model that utilizes data from IMUs and UWB sensors. The performance of the proposed solution has been verified through simulation and experimental results, showcasing that the proposed solution is able to provide a sensibly accurate 3D point cloud representation in real-world agricultural environments, marking it as a promising solution in precision agriculture.