Evaluation of LiDAR-based row following control for UAS precision spraying

In precision agriculture, automation is playing a key role in revolutionizing crop monitoring tasks and in-field operations. In this context, precision and spot spraying, performed by Unmanned Aerial Systems (UASs), are of great potential, allowing to save chemicals reducing drift, and modulating the spray in accordance with the real crop need. In addition, the UAS spraying can be performed when and/or where ground vehicles cannot be adopted. In this work, the feasibility study of exploiting an on-board LiDAR based control system has been conducted, which allows UASs to precisely follow the vine row. In particular, a simulation tool has been developed to evaluate the relation between the altitude and pose of UAS and the LiDAR sensor Field Of View (FOV) within a detailed 3D model of vineyards. Moreover, the adoption of several LiDAR model can be simulated. The simulation tool was validated with real data from an experimental campaign held at DiSAFA vineyard in September 2023: (i) the accurate 3D model of the vineyard was generated by RGB imagery acquired with a DJI Mavic 2 Pro, and (ii) about 10,000 DJI Zenmuse L1 LiDAR frames have been acquired, using a DJI Matrice 350 RTK. Results proved that the LiDAR frame generated by the simulation tool are consistent with reference in-field acquired one. The developed simulation tool enables the optimisation of the row following control system: indeed, different LiDAR models and mounting angles, and several vine row detection algorithms, can be tested avoiding expensive hardware implementation and in-field experiments.