Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

SafeLand: Safe Autonomous Landing in Unknown Environments with Bayesian Semantic Mapping

Markus Gross, Andreas Greiner, Sai Bharadhwaj Matha, Felix Soest, Daniel Cremers, Henri Meeß

Abstract

Autonomous landing of uncrewed aerial vehicles (UAVs) in unknown, dynamic environments poses significant safety challenges, particularly near people and infrastructure, as UAVs transition to routine urban and rural operations. Existing methods often rely on prior maps, heavy sensors like LiDAR, static markers, or fail to handle non-cooperative dynamic obstacles like humans, limiting generalization and real-time performance. To address these challenges, we introduce SafeLand, a lean, vision-based system for safe autonomous landing (SAL) that requires no prior information and operates only with a camera and a lightweight height sensor. Our approach constructs an online semantic ground map via deep learning-based semantic segmentation, optimized for embedded deployment and trained on a consolidation of seven curated public aerial datasets (achieving 70.22% mIoU across 20 classes), which is further refined through Bayesian probabilistic filtering with temporal semantic decay to robustly identify metric-scale landing spots. A behavior tree then governs adaptive landing, iteratively validates the spot, and reacts in real time to dynamic obstacles by pausing, climbing, or rerouting to alternative spots, maximizing human safety. We extensively evaluate our method in 200 simulations and 60 end-to-end field tests across industrial, urban, and rural environments at altitudes up to 100m, demonstrating zero false negatives for human detection. Compared to the state of the art, SafeLand achieves sub-second response latency, substantially lower than previous methods, while maintaining a superior success rate of 95%. To facilitate further research in aerial robotics, we release SafeLand's segmentation model as a plug-and-play ROS package, available at https://github.com/markus-42/SafeLand.

SafeLand: Safe Autonomous Landing in Unknown Environments with Bayesian Semantic Mapping

Abstract

Autonomous landing of uncrewed aerial vehicles (UAVs) in unknown, dynamic environments poses significant safety challenges, particularly near people and infrastructure, as UAVs transition to routine urban and rural operations. Existing methods often rely on prior maps, heavy sensors like LiDAR, static markers, or fail to handle non-cooperative dynamic obstacles like humans, limiting generalization and real-time performance. To address these challenges, we introduce SafeLand, a lean, vision-based system for safe autonomous landing (SAL) that requires no prior information and operates only with a camera and a lightweight height sensor. Our approach constructs an online semantic ground map via deep learning-based semantic segmentation, optimized for embedded deployment and trained on a consolidation of seven curated public aerial datasets (achieving 70.22% mIoU across 20 classes), which is further refined through Bayesian probabilistic filtering with temporal semantic decay to robustly identify metric-scale landing spots. A behavior tree then governs adaptive landing, iteratively validates the spot, and reacts in real time to dynamic obstacles by pausing, climbing, or rerouting to alternative spots, maximizing human safety. We extensively evaluate our method in 200 simulations and 60 end-to-end field tests across industrial, urban, and rural environments at altitudes up to 100m, demonstrating zero false negatives for human detection. Compared to the state of the art, SafeLand achieves sub-second response latency, substantially lower than previous methods, while maintaining a superior success rate of 95%. To facilitate further research in aerial robotics, we release SafeLand's segmentation model as a plug-and-play ROS package, available at https://github.com/markus-42/SafeLand.
Paper Structure (14 sections, 9 equations, 6 figures, 2 tables)

This paper contains 14 sections, 9 equations, 6 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Methodology
  4. Method Overview
  5. Image Acquisition and Semantic Segmentation
  6. Projection and Registration
  7. Bayesian Filtering and Semantic Decay
  8. Landing Spot Detection
  9. Landing Procedure
  10. Experimental Results
  11. Simulation Experiments
  12. Field Experiments
  13. Baseline Comparison
  14. Conclusion

Figures (6)

  • Figure 1: SafeLand predicts a semantic map, projects it to a metric ground plane, filters it probabilistically over time to select a landing spot, and uses a behavior tree for safe real-time landing in unknown dynamic environments.
  • Figure 2: Chronological processing steps of our method SafeLand. An overview is provided in Sec. \ref{['subsec_method_overview']}.
  • Figure 3: Landing procedure. A behavior tree (Sec. \ref{['subsec_landing_logic']}) executes three distinct sequences that ensure safe, real-time landing in unknown and dynamic environments.
  • Figure 4: Hardware configuration. Left: Quadcopter PM Q-685 premo custom frame, equipped with NVIDIA Jetson AGX Orin orin, GNSS module SimpleRTK2B Pro gnss, and flight controller Pixhawk Cube Orange with its IMUs pixhwak. Right: Nadir-pointing RGB camera FLIR Blackfly S-PGE-31S4C-C flir and the height AGL sensor Lightware SF11/C rangefinder.
  • Figure 5: System response latencies (see Sec. \ref{['subsec_comparison']}) on dynamic obstacles for Safe2Ditch safe2ditch and SafeLand (ours). We substantially outperform Safe2Ditch by providing sub-second response as a critical capability for unknown environments.
  • ...and 1 more figures