ForestVO: Enhancing Visual Odometry in Forest Environments through ForestGlue

2025-04-02

AI Summary

Visual odometry in forest environments suffers from poor feature matching robustness and low pose estimation accuracy due to dense vegetation, highly variable illumination, and repetitive textures. Method: This paper proposes ForestGlue—a multimodal feature extraction and matching framework—and a lightweight visual odometry system, ForestVO. ForestGlue introduces the first forest-specific feature detector, supporting RGB, RGB-D, grayscale, and stereo inputs. It integrates an enhanced SuperPoint architecture, retrained LightGlue/SuperGlue matchers, Transformer-based pose estimation, and synthetic forest data augmentation. A forest-tailored end-to-end training paradigm is adopted. Contribution/Results: Using only 10% annotated data, ForestVO achieves state-of-the-art performance: relative pose error (RPE) of 1.09 m on TartanAir forest sequences and KITTI-score of 2.33%—a 40% improvement over DSO. It reduces computational overhead by 75% and keypoint count by 75%, while maintaining baseline accuracy.

Technology Category

• Computer Vision: Multi-modal Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Machine Learning: Multimodal Learning

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Recent advancements in visual odometry systems have improved autonomous navigation; however, challenges persist in complex environments like forests, where dense foliage, variable lighting, and repetitive textures compromise feature correspondence accuracy. To address these challenges, we introduce ForestGlue, enhancing the SuperPoint feature detector through four configurations - grayscale, RGB, RGB-D, and stereo-vision - optimised for various sensing modalities. For feature matching, we employ LightGlue or SuperGlue, retrained with synthetic forest data. ForestGlue achieves comparable pose estimation accuracy to baseline models but requires only 512 keypoints - just 25% of the baseline's 2048 - to reach an LO-RANSAC AUC score of 0.745 at a 10{deg} threshold. With only a quarter of keypoints needed, ForestGlue significantly reduces computational overhead, demonstrating effectiveness in dynamic forest environments, and making it suitable for real-time deployment on resource-constrained platforms. By combining ForestGlue with a transformer-based pose estimation model, we propose ForestVO, which estimates relative camera poses using matched 2D pixel coordinates between frames. On challenging TartanAir forest sequences, ForestVO achieves an average relative pose error (RPE) of 1.09 m and a kitti_score of 2.33%, outperforming direct-based methods like DSO by 40% in dynamic scenes. Despite using only 10% of the dataset for training, ForestVO maintains competitive performance with TartanVO while being a significantly lighter model. This work establishes an end-to-end deep learning pipeline specifically tailored for visual odometry in forested environments, leveraging forest-specific training data to optimise feature correspondence and pose estimation, thereby enhancing the accuracy and robustness of autonomous navigation systems.

Problem

• Improving visual odometry accuracy in complex forest environments
• Reducing computational overhead with fewer keypoints for real-time deployment
• Enhancing feature correspondence and pose estimation in dynamic forest scenes

Innovation

• ForestGlue enhances SuperPoint with four configurations
• LightGlue/SuperGlue retrained with synthetic forest data
• Transformer-based pose estimation model for ForestVO