UAV Coverage Path Planner

A standalone Python library and CLI for constraint-aware UAV area-coverage route planning. It creates lawnmower routes inside a polygon and uses a deterministic genetic algorithm to jointly optimize:

• heading angle;
• lane spacing;
• cruise speed.

The optimizer derives feasible lane-spacing and speed bounds from generic camera geometry, desired ground sampling distance (GSD), overlap requirements, capture interval, aircraft speed, range, flight time and reserve energy. It contains no maps, hardware drivers, user data, model files, branding assets, or application code.

中文说明见 README_zh-CN.md。

Install

python -m pip install uav-coverage-path-planner

For development:

git clone https://github.com/luoyuejun9/uav-coverage-path-planner.git
cd uav-coverage-path-planner
python -m pip install -e ".[dev]"
pytest

Quick start

from uav_coverage_path_planner import PlanningProblem, optimize_coverage_route

problem = PlanningProblem(
    boundary=((0, 0), (160, 10), (180, 110), (95, 145), (10, 105)),
    takeoff_point=(-25, -30),
    task_height_m=55,
)
result = optimize_coverage_route(problem)

print(result.optimized.heading_deg)
print(result.optimized.lane_spacing_m)
print(result.optimized.cruise_speed_mps)

local_xy is the default coordinate system and uses metres. Set coordinate_system="wgs84" to pass (longitude, latitude) points; the library internally applies a local equirectangular projection and returns the route in WGS84 again.

CLI

uav-route-optimize examples/synthetic_mission.json --output output

This writes only runtime results to output/:

• result.json — full optimization result;
• route.geojson — coverage polyline;
• convergence.csv — generation history;
• population.csv — final population sample;
• analysis.png — four-panel diagnostic chart.

The included mission is a synthetic local-metre example. It contains no real locations.

Model

For a chromosome x = [heading, lane_spacing, cruise_speed], the algorithm minimizes:

J(x) = w_geometry × J_geometry + w_time × J_time + w_energy × J_energy + w_quality × J_quality + w_speed × J_speed

subject to GSD, side-overlap, forward-overlap, capture-interval, speed, range and reserve-energy constraints. Infeasible individuals are ranked behind feasible individuals and carry a violation penalty. The default population is 50, with two deterministic restarts and early stopping after stagnant generations.

fine, balanced, efficient, and custom profiles only change objective weights and quality constraints supplied by the caller; they do not encode a specific aircraft or commercial platform.

Reproducibility

Pass OptimizationConfig(seed=...) for a repeatable run. If no seed is supplied, a stable seed is derived from the mission definition.

Safety note

This package is a planning aid, not a flight controller. Validate all routes, geofencing, terrain clearance, communications, local aviation rules and aircraft limits before operating a UAV.

License

MIT. See LICENSE.