Table of Contents
Fetching ...

AeroTraj: Trajectory Planning for Fast, and Accurate 3D Reconstruction Using a Drone-based LiDAR

Fawad Ahmad, Christina Shin, Rajrup Ghosh, John D'Ambrosio, Eugene Chai, Karthik Sundaresan, Ramesh Govindan

Abstract

This paper presents AeroTraj, a system that enables fast, accurate, and automated reconstruction of 3D models of large buildings using a drone-mounted LiDAR. LiDAR point clouds can be used directly to assemble 3D models if their positions are accurately determined. AeroTraj uses SLAM for this, but must ensure complete and accurate reconstruction while minimizing drone battery usage. Doing this requires balancing competing constraints: drone speed, height, and orientation. AeroTraj exploits building geometry in designing an optimal trajectory that incorporates these constraints. Even with an optimal trajectory, SLAM's position error can drift over time, so AeroTraj tracks drift in-flight by offloading computations to the cloud and invokes a re-calibration procedure to minimize error. AeroTraj can reconstruct large structures with centimeter-level accuracy and with an average end-to-end latency below 250 ms, significantly outperforming the state of the art.

AeroTraj: Trajectory Planning for Fast, and Accurate 3D Reconstruction Using a Drone-based LiDAR

Abstract

This paper presents AeroTraj, a system that enables fast, accurate, and automated reconstruction of 3D models of large buildings using a drone-mounted LiDAR. LiDAR point clouds can be used directly to assemble 3D models if their positions are accurately determined. AeroTraj uses SLAM for this, but must ensure complete and accurate reconstruction while minimizing drone battery usage. Doing this requires balancing competing constraints: drone speed, height, and orientation. AeroTraj exploits building geometry in designing an optimal trajectory that incorporates these constraints. Even with an optimal trajectory, SLAM's position error can drift over time, so AeroTraj tracks drift in-flight by offloading computations to the cloud and invokes a re-calibration procedure to minimize error. AeroTraj can reconstruct large structures with centimeter-level accuracy and with an average end-to-end latency below 250 ms, significantly outperforming the state of the art.

Paper Structure

This paper contains 43 sections, 2 equations, 18 figures, 15 tables, 1 algorithm.

Figures (18)

  • Figure 1: Compared to a ground truth 3D model (a), an MVS/photogrammetry-based 3D reconstruction (b) of a building can be incompletei.e., it contains holes and missing regions.
  • Figure 2: Three models of a large complex on our campus: GPS (a), SLAM (b), and AeroTraj (c). All models use the same height ramp function to color-code the z-axis, with blue representing the lowest points and orange representing the highest. The AeroTraj model has distinct features and trees, with clear and crisp coloring, indicating a good reconstruction. In contrast, the GPS and SLAM models have diffused colors and lack visible features, suggesting poor and noisy reconstruction.
  • Figure 3: LiDAR returns when mounted on a vehicle (orange lines) and a drone platform (blue, and green lines). Solid lines represent the laser beams which hit other objects (non-zero returns), whereas the dotted lines represent ones that did not (zero returns).
  • Figure 4: Rectilinear trajectory for SLAM-based reconstruction.
  • Figure 5: AeroTraj is designed for buildings with vertical sides and either polygonal or gabled roofs and their variations (hipped, mansard, pyramidal, skillion, etc.).
  • ...and 13 more figures