Pathfinder for Low-altitude Aircraft with Binary Neural Network

2024-09-13

arXiv.org

AI Summary

To address the incompleteness of prior topological maps—such as OpenStreetMap (OSM)—for ground robot autonomous mapping, caused by missing road annotations, this paper proposes a high-precision road skeleton extraction method tailored for low-altitude aerial platforms. Our approach fuses airborne LiDAR point clouds and RGB imagery within an end-to-end binary dual-stream road segmentation framework. We design an attention-guided gated module to adaptively fuse sparse point cloud and image features, integrate a binarized Vision Transformer (ViT) encoder, and introduce a focal-aware joint loss function. Evaluated on two public benchmarks, our method achieves state-of-the-art segmentation accuracy while significantly reducing model parameters and computational cost—enabling efficient edge deployment. The resulting road prior map exhibits structural completeness and topological fidelity comparable to OSM, thereby robustly supporting ground robot mapping in real-world environments.

Technology Category

• Computer Vision: Segmentation
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Planning, Routing, and Scheduling: Planning with Language Models

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

A prior global topological map (e.g., the OpenStreetMap, OSM) can boost the performance of autonomous mapping by a ground mobile robot. However, the prior map is usually incomplete due to lacking labeling in partial paths. To solve this problem, this paper proposes an OSM maker using airborne sensors carried by low-altitude aircraft, where the core of the OSM maker is a novel efficient pathfinder approach based on LiDAR and camera data, i.e., a binary dual-stream road segmentation model. Specifically, a multi-scale feature extraction based on the UNet architecture is implemented for images and point clouds. To reduce the effect caused by the sparsity of point cloud, an attention-guided gated block is designed to integrate image and point-cloud features. To optimize the model for edge deployment that significantly reduces storage footprint and computational demands, we propose a binarization streamline to each model component, including a variant of vision transformer (ViT) architecture as the encoder of the image branch, and new focal and perception losses to optimize the model training. The experimental results on two datasets demonstrate that our pathfinder method achieves SOTA accuracy with high efficiency in finding paths from the low-level airborne sensors, and we can create complete OSM prior maps based on the segmented road skeletons. Code and data are available at: href{https://github.com/IMRL/Pathfinder}{https://github.com/IMRL/Pathfinder}.

Problem

• Incomplete prior maps hinder autonomous robot navigation.
• Proposes a binary neural network for efficient pathfinding.
• Optimizes model for edge deployment with reduced resources.

Innovation

• Binary dual-stream road segmentation model
• Attention-guided gated block integration
• Binarization streamline for edge deployment