Active Semantic Mapping of Horticultural Environments Using Gaussian Splatting
🤖 AI Summary
This work addresses the inefficiency and poor scalability of traditional agricultural semantic reconstruction, which often relies on manual labor or fixed cameras. The authors propose an active 3D reconstruction framework leveraging a mobile manipulator, integrating a low-resolution Octomap for efficient viewpoint selection and motion planning, and employing 3D Gaussian Splatting to achieve high-fidelity semantic mapping. By incorporating a segmentation-noise-robust mechanism and memory-efficient strategies, the system significantly improves both accuracy and computational efficiency. In simulated environments, it achieves up to a 28.6% increase in fruit-level F1 score and a 50% reduction in runtime compared to conventional occupancy-grid-based approaches, demonstrating a strong balance between precision and performance.
Application Category
📝 Abstract
Semantic reconstruction of agricultural scenes plays a vital role in tasks such as phenotyping and yield estimation. However, traditional approaches that rely on manual scanning or fixed camera setups remain a major bottleneck in this process. In this work, we propose an active 3D reconstruction framework for horticultural environments using a mobile manipulator. The proposed system integrates the classical Octomap representation with 3D Gaussian Splatting to enable accurate and efficient target-aware mapping. While a low-resolution Octomap provides probabilistic occupancy information for informative viewpoint selection and collision-free planning, 3D Gaussian Splatting leverages geometric, photometric, and semantic information to optimize a set of 3D Gaussians for high-fidelity scene reconstruction. We further introduce simple yet effective strategies to enhance robustness against segmentation noise and reduce memory consumption. Simulation experiments demonstrate that our method outperforms purely occupancy-based approaches in both runtime efficiency and reconstruction accuracy, enabling precise fruit counting and volume estimation. Compared to a 0.01m-resolution Octomap, our approach achieves an improvement of 6.6% in fruit-level F1 score under noise-free conditions, and up to 28.6% under segmentation noise. Additionally, it achieves a 50% reduction in runtime, highlighting its potential for scalable, real-time semantic reconstruction in agricultural robotics.
Problem
Research questions and friction points this paper is trying to address.
semantic reconstruction
horticultural environments
3D mapping
agricultural robotics
fruit phenotyping
Innovation
Methods, ideas, or system contributions that make the work stand out.
Active 3D reconstruction
3D Gaussian Splatting
Semantic mapping
Agricultural robotics
Octomap
🔎 Similar Papers
No similar papers found.
