Neural 3D Object Reconstruction with Small-Scale Unmanned Aerial Vehicles

2025-09-15

AI Summary

Micro-scale UAVs (e.g., Crazyflie 2.1) face fundamental limitations in payload capacity and autonomy, hindering high-fidelity 3D reconstruction. Method: We propose a fully autonomous neuro-geometric co-reconstruction framework featuring a dual-pipeline architecture that tightly integrates ultra-wideband (UWB) localization, structure-from-motion (SfM), and neural radiance fields (NeRF). This enables a real-time closed-loop between data acquisition and flight control: near-real-time point cloud analysis dynamically optimizes flight trajectories for adaptive spatial coverage. Results: Experiments demonstrate substantial improvements over static path planning in confined and unstructured environments. Both single- and multi-UAV configurations achieve high-resolution, geometrically complete 3D reconstructions. To our knowledge, this is the first end-to-end autonomous modeling system on a centimeter-scale platform that jointly ensures geometric accuracy and photorealistic appearance fidelity.

Technology Category

• Intelligent Robots: Localization, Mapping, and Navigation
• Computer Vision: 3D Computer Vision
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Small Unmanned Aerial Vehicles (UAVs) exhibit immense potential for navigating indoor and hard-to-reach areas, yet their significant constraints in payload and autonomy have largely prevented their use for complex tasks like high-quality 3-Dimensional (3D) reconstruction. To overcome this challenge, we introduce a novel system architecture that enables fully autonomous, high-fidelity 3D scanning of static objects using UAVs weighing under 100 grams. Our core innovation lies in a dual-reconstruction pipeline that creates a real-time feedback loop between data capture and flight control. A near-real-time (near-RT) process uses Structure from Motion (SfM) to generate an instantaneous pointcloud of the object. The system analyzes the model quality on the fly and dynamically adapts the UAV's trajectory to intelligently capture new images of poorly covered areas. This ensures comprehensive data acquisition. For the final, detailed output, a non-real-time (non-RT) pipeline employs a Neural Radiance Fields (NeRF)-based Neural 3D Reconstruction (N3DR) approach, fusing SfM-derived camera poses with precise Ultra Wide-Band (UWB) location data to achieve superior accuracy. We implemented and validated this architecture using Crazyflie 2.1 UAVs. Our experiments, conducted in both single- and multi-UAV configurations, conclusively show that dynamic trajectory adaptation consistently improves reconstruction quality over static flight paths. This work demonstrates a scalable and autonomous solution that unlocks the potential of miniaturized UAVs for fine-grained 3D reconstruction in constrained environments, a capability previously limited to much larger platforms.

Problem

• Enabling high-fidelity 3D reconstruction with lightweight UAVs under 100g
• Overcoming payload and autonomy constraints for complex scanning tasks
• Creating real-time feedback between data capture and flight control

Innovation

• Dual-reconstruction pipeline for real-time feedback
• Dynamic trajectory adaptation for comprehensive coverage
• NeRF-based fusion with SfM and UWB data