ForaNav: Insect-inspired Online Target-oriented Navigation for MAVs in Tree Plantations

TL;DR

The paper addresses GPS-denied navigation for MAVs in plantations by introducing ForaNav, an online target-oriented navigation framework. It integrates a real-time oil palm detection pipeline based on an enhanced hierarchical HOG that leverages hue-saturation histograms and global HOG variance with a Kalman-filter–driven, insect-inspired navigation loop and a view-memory recovery mechanism. The approach generalizes to other tree types, achieves favorable real-time performance on a lightweight onboard platform, and demonstrates precise targeting with mean trajectory deviation under $0.1\,m$ across diverse layouts without prior tree locations. Its practical impact lies in enabling autonomous precision agriculture tasks like canopy assessment and targeted interventions in GPS-challenged rural plantations.

Abstract

Autonomous Micro Air Vehicles (MAVs) are becoming essential in precision agriculture to enhance efficiency and reduce labor costs through targeted, real-time operations. However, existing unmanned systems often rely on GPS-based navigation, which is prone to inaccuracies in rural areas and limits flight paths to predefined routes, resulting in operational inefficiencies. To address these challenges, this paper presents ForaNav, an insect-inspired navigation strategy for autonomous navigation in plantations. The proposed method employs an enhanced Histogram of Oriented Gradient (HOG)-based tree detection approach, integrating hue-saturation histograms and global HOG feature variance with hierarchical HOG extraction to distinguish oil palm trees from visually similar objects. Inspired by insect foraging behavior, the MAV dynamically adjusts its path based on detected trees and employs a recovery mechanism to stay on course if a target is temporarily lost. We demonstrate that our detection method generalizes well to different tree types while maintaining lower CPU usage, lower temperature, and higher FPS than lightweight deep learning models, making it well-suited for real-time applications. Flight test results across diverse real-world scenarios show that the MAV successfully detects and approaches all trees without prior tree location, validating its effectiveness for agricultural automation.

Figures (9)

• Figure 1: Schematic of insect-inspired egocentric visual navigation. Inspired by the foraging behavior of insects, the MAV navigates with targeting oil palm trees. The MAV detects trees in real time using an enhanced HOG-based method. Their image coordinates $p_{t}$ are tracked with a Kalman filter to estimate $\hat{p}_{t}$ which is used to update MAV trajectory online to approach each target.
• Figure 2: Overview of the detection and navigation frameworks. The onboard camera system captures images and detects trees in real time, using their image coordinates to update the MAV trajectory via the path planning module and track targets through the motion control module.
• Figure 3: Workflow diagram of the proposed oil palm tree detection method. The detection window is classified as an oil palm tree based on SVM classification, variance calculations, and hue and saturation comparisons.
• Figure 4: Schematic diagram of the hierarchical HOG feature extraction of oil palm trees. The green arrows represent the HOG features across different layers. The orange arrows depict the star-shaped HOG features of pinnate leaves. The purple arrows show the needle-like structures of leaflets.
• Figure 5: Flight test results in Map 1. Flight trajectories for Tests 1-3 are plotted in the center. Sample detection results from Test 1 are showed on the right.