AgriGS-SLAM: Orchard Mapping Across Seasons via Multi-View Gaussian Splatting SLAM

2025-10-30

AI Summary

Cross-season 3D mapping in orchards remains challenging due to repetitive row structures, drastic appearance changes across seasons, and dynamic occlusions (e.g., wind-induced foliage motion), severely degrading SLAM robustness. Method: We propose a multimodal Gaussian point-lattice SLAM system that fuses visual and LiDAR data. A gradient-driven map lifecycle mechanism is introduced, alongside LiDAR depth-consistency priors for pose optimization to strengthen geometry-appearance coupling. Direct LiDAR odometry and loop closure are integrated with multi-camera 3D Gaussian rendering and batch rasterization to recover structure under occlusion. Results: Evaluated on apple and pear orchards, our system significantly improves cross-season reconstruction stability and trajectory accuracy, enables novel-view synthesis for quantitative assessment, and maintains real-time performance on-board—outperforming state-of-the-art methods.

Technology Category

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

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Autonomous robots in orchards require real-time 3D scene understanding despite repetitive row geometry, seasonal appearance changes, and wind-driven foliage motion. We present AgriGS-SLAM, a Visual--LiDAR SLAM framework that couples direct LiDAR odometry and loop closures with multi-camera 3D Gaussian Splatting (3DGS) rendering. Batch rasterization across complementary viewpoints recovers orchard structure under occlusions, while a unified gradient-driven map lifecycle executed between keyframes preserves fine details and bounds memory. Pose refinement is guided by a probabilistic LiDAR-based depth consistency term, back-propagated through the camera projection to tighten geometry-appearance coupling. We deploy the system on a field platform in apple and pear orchards across dormancy, flowering, and harvesting, using a standardized trajectory protocol that evaluates both training-view and novel-view synthesis to reduce 3DGS overfitting in evaluation. Across seasons and sites, AgriGS-SLAM delivers sharper, more stable reconstructions and steadier trajectories than recent state-of-the-art 3DGS-SLAM baselines while maintaining real-time performance on-tractor. While demonstrated in orchard monitoring, the approach can be applied to other outdoor domains requiring robust multimodal perception.

Problem

• Mapping orchards across seasons with real-time 3D reconstruction
• Handling repetitive geometry and seasonal appearance changes
• Improving SLAM accuracy under occlusion and foliage motion

Innovation

• Multi-camera Gaussian Splatting for real-time 3D reconstruction
• Unified gradient-driven map lifecycle preserving fine details
• Probabilistic LiDAR depth consistency tightening geometry-appearance coupling