Autonomous navigation for low-altitude UAVs in urban areas

Thomas Castelli, Aidean Sharghi, Don Harper, Alain Trémeau, Mubarak ShaharXiv 2016 · OpenReview

Abstract

In recent years, consumer Unmanned Aerial Vehicles (UAVs) have become ubiquitous, allowing inexperienced users to operate drones in complex environments. This raises significant concerns regarding public safety and regulatory compliance. This paper presents a novel approach to enable safe UAV operation in urban areas by leveraging geodetically accurate satellite imagery and Geographical Information System (GIS) data. Our method utilizes a weighted A* algorithm to compute flight paths that prioritize public safety by minimizing the risk of injury to individuals on the ground. By categorizing land-use risk (e.g., flying over buildings is safer than over roads) and implementing real-time dynamic avoidance of moving objects, we demonstrate a navigation system that is over 40 times safer than traditional straight-line methods.

Introduction

As consumer drones become a common sight in urban skies, the risk of mechanical failure or pilot error leading to ground-level injuries increases. Most current autopilot systems rely on simple straight-line navigation between GPS waypoints, which ignores the underlying environment. A crash over a busy thoroughfare is far more likely to cause harm than a crash over an unoccupied rooftop. This research addresses the need for "environment-aware" navigation that automates flight paths to minimize human risk.

Methodology

The proposed system employs a two-tier navigation strategy to manage both static environmental risks and dynamic threats.

Global Path Planning (Static)

The system integrates satellite imagery with GIS data (via Google Maps API) to construct a multi-layered risk map. Land-use categories are assigned specific weights based on the probability of human presence:

  • High Risk (Roads/Paths): Weight of 100.
  • Low Risk (Buildings/Water): Weight of 5 (acting as a safety buffer).
  • Neutral (Other): Weight of 20.

Using these weights, the A* algorithm calculates a global path that minimizes the total risk cost rather than just the physical distance.

Figure 2: The multi-layered approach integrating GIS road networks, building masks, and satellite imagery to create a comprehensive risk map.

Local Path Planning (Dynamic)

To account for moving objects like cars and pedestrians, the drone utilizes its onboard camera and computer vision. For every detected moving object, the system projects a "cost bubble"—a multivariate normal distribution—onto the map. This bubble is dynamically shaped by the object's velocity and predicted trajectory. If the drone's path intersects with a high-cost bubble, the local path is recalculated in real-time to maintain a safe standoff distance.

Figure 6: Real-time dynamic avoidance showing the UAV (airplane icon) adjusting its path (blue line) to steer clear of moving vehicles and their associated risk heatmaps.

Key Insights

  • Counter-Intuitive Safety: A core finding is that flying over buildings is significantly safer than flying over roads. Buildings provide a physical barrier that protects people inside, whereas roads and sidewalks are high-exposure zones.
  • Geo-Registration: By precisely aligning real-time video frames with static satellite maps, the system treats pixels and GPS coordinates as interchangeable, allowing visual maneuvers to be translated directly into autopilot waypoints.
  • Predictive Risk: The use of velocity-stretched "risk bubbles" allows the system to plan around where a threat will be, rather than where it is currently located.

Results and Discussion

The system was validated through 54 simulated missions using Wide-Area Motion Imagery (WAMI) datasets (WPAFB and PVLabs).

  • Safety Performance: The dynamic navigation method achieved a safety parameter over 40 times better than naive straight-line navigation.
  • Efficiency Trade-off: Safety comes at a minor cost to flight efficiency. The risk-aware paths were 6% to 20% longer than straight lines, adding between 32 seconds and 5 minutes to typical missions—a negligible increase given the substantial reduction in liability.

Conclusion

This research demonstrates that environment-aware navigation is a viable and necessary step for the safe integration of UAVs into urban airspace. While the current model relies on GIS accuracy and high-quality imagery, future work will focus on real-time onboard processing and the detection of mid-air obstacles like power lines and trees to create a fully autonomous, safety-first navigation suite.

Summary updated

Abstract

In recent years, consumer Unmanned Aerial Vehicles have become very popular, everyone can buy and fly a drone without previous experience, which raises concern in regards to regulations and public safety. In this paper, we present a novel approach towards enabling safe operation of such vehicles in urban areas. Our method uses geodetically accurate dataset images with Geographical Information System (GIS) data of road networks and buildings provided by Google Maps, to compute a weighted A* shortest path from start to end locations of a mission. Weights represent the potential risk of injuries for individuals in all categories of land-use, i.e. flying over buildings is considered safer than above roads. We enable safe UAV operation in regards to 1- land-use by computing a static global path dependent on environmental structures, and 2- avoiding flying over moving objects such as cars and pedestrians by dynamically optimizing the path locally during the flight. As all input sources are first geo-registered, pixels and GPS coordinates are equivalent, it therefore allows us to generate an automated and user-friendly mission with GPS waypoints readable by consumer drones' autopilots. We simulated 54 missions and show significant improvement in maximizing UAV's standoff distance to moving objects with a quantified safety parameter over 40 times better than the naive straight line navigation.
BibTeX
@misc{castelli2016autonomous,
  author = {Thomas Castelli and Aidean Sharghi and Don Harper and Alain Trémeau and Mubarak Shah},
  title = {{Autonomous navigation for low-altitude UAVs in urban areas}},
  year = {2016},
  eprint = {1602.08141},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1602.08141},
}
Figure