ARCHIVES
VOL. 1, ISSUE 2 (2025)
Autonomous navigation algorithms for agricultural drones in dense canopies: A comparative study of LiDAR-fusion and vision-based approaches
Authors
Kaviya Jeyabalan
Abstract
Autonomous navigation
in dense agricultural canopies remains a critical challenge for precision
farming applications. Traditional Global Navigation Satellite System (GNSS)
based approaches fail beneath dense foliage, necessitating alternative sensing
and navigation strategies. This study evaluated three autonomous navigation
algorithms—LiDAR-Fusion Extended Kalman Filter (LF-EKF), Monocular Vision-based
Visual Simultaneous Localization and Mapping (V-SLAM), and a hybrid approach
combining both modalities (Hybrid-Fusion)—for navigating aerial drones through
representative dense apple and grape vineyard canopies. Field experiments were
conducted across six test sites covering approximately 2.4 hectares of
horticultural crop environments during the 2023–2024 growing seasons. Each
algorithm was evaluated based on positional accuracy, trajectory smoothness,
computational efficiency, and operational success rate. Results demonstrated
that the Hybrid-Fusion approach achieved the highest positional accuracy with root
mean square error (RMSE) of 0.18 m and 89% mission success rate, compared to
0.34 m and 71% for V-SLAM and 0.41 m and 63% for LF-EKF. The Hybrid-Fusion
algorithm exhibited superior robustness to occlusions and rapid lighting
changes characteristic of dense canopy environments. Mean computational load
was 1.8 cores on a quad-core onboard processor, with average power consumption
of 18.4 W. The proposed hybrid framework demonstrates substantial improvement
over single-modality approaches and exhibits commercial viability for precision
agricultural applications such as selective harvesting, phytosanitary
monitoring, and targeted pesticide application. These findings have significant
implications for enabling autonomous agricultural drones to operate effectively
in challenging visibility conditions, potentially increasing operational
efficiency by 62–75% compared to manual inspection methods. Future developments
should focus on expanding algorithm validation across diverse global crop types
and investigating edge-computing optimization for extended flight endurance.
Download
Pages:1-10
How to cite this article:
Kaviya Jeyabalan "Autonomous navigation algorithms for agricultural drones in dense canopies: A comparative study of LiDAR-fusion and vision-based approaches". World Journal of Engineering and Technology, Vol 1, Issue 2, 2025, Pages 1-10
Download Author Certificate
Please enter the email address corresponding to this article submission to download your certificate.