Robotics Research Hub
Safe Aerial 3D Path Planning for Autonomous UAVs using Magnetic Potential Fields

Safe Aerial 3D Path Planning for Autonomous UAVs using Magnetic Potential Fields

Haechan Mark Bong, Giovanni Beltrame · · 2026

uav navigation
PDF

Framework

N/A

License

N/A

Stars

0

Summary

Safe autonomous Uncrewed Aerial Vehicle (UAV) navigation in urban environments requires real-time path planning that avoids obstacles. MaxConvNet is a potential-field planner that leverages properties of Maxwell's equations to generate a path to the goal without local minima.

Abstract Summary

Safe autonomous navigation in three-dimensional environments remains a critical challenge for unmanned aerial vehicles (UAVs). This paper proposes a novel path planning approach based on magnetic potential fields that provides collision-free trajectories while maintaining computational efficiency.... The method demonstrates significant improvements over existing approaches, providing both theoretical insights and practical benefits for real-world deployment. Comprehensive experiments validate the effectiveness of the proposed approach across diverse scenarios and task settings.

Key Points

  • Proposes Safe Aerial 3D Path Planning for Autonomous UAVs using Magnetic Potential Fields, a novel approach for uav in robotics.
  • Addresses key limitations in existing methods through innovative architecture design.
  • Demonstrates strong empirical results on standard benchmarks and real-world evaluations.
  • Introduces novel training methodology that improves generalization and sample efficiency.
  • Provides comprehensive analysis of failure modes and ablation studies.

Method

The environment is modeled as an electromagnetic conduction problem on a $101^{3}$ voxel grid ($\pm$100m, 2m resolution).

Obstacle voxels (from LiDAR, dilated by 4 voxels for safety margin) have zero conductivity ($\sigma_{o}\!=\!0$), the goal has high conductivity ($\sigma_{g}\!=\!10^{6}$), and free space combines initial conductivity ($\sigma_{i}\!=\!10^{2}$) to determine the distance-weighted conductivity as defefined below: We extend MaxConvNet [ 9 ] from $N^{2}$ grids to $101^{3}$ volumes.

Method — Safe Aerial 3D Path Planning for Autonomous UAVs using Magnetic Potential Fields
Fig. 4: Field visualizations using Cosys-AirSim [ 15 ] , with LA Night City [ 1 ] (top) and London White City [ 2 ] (bottom), extracted from experiment 45. Conductivity: The raw conductivity field ($\sigma$) and its log-scaled representation ($\sigma_{\log}$) showing the distance-weighted gradient from free space toward the goal. Magnetic field: The 3DMaxConvNet predicted magnetic field ($\phi$) and its normalized form ($\phi_{\mathrm{normalized}}$) revealing the gradient structure that guides the UAV toward the goal. The white and red sphere marks the start, the goal, respectively and the planned path follows the field gradient.

Experiments

We evaluate in Cosys-AirSim [ 15 ] on two urban environments (Fig. 4 ): LA Night City [ 1 ] is a dense urban environment, mirroring the real LA city by featuring high-rise buildings, narrow streets, and night-time lighting conditions. The complex geometry with tall structures and confined passages tests the planner’s ability to navigate through cluttered 3D obstacle fields.

London White City [ 2 ] is a suburban district with low-rise buildings and open spaces.

Sources and Demonstrations

Figures are reproduced from Bong et al. See the full paper for experimental details and the project page, when the paper links one, for demonstrations and videos.

Share

Related Papers