Cooperative Multi-UAV Coverage Mission Planning Platform for Remote Sensing Applications - Path Planning back-end

Description

The project in this repository is a multi-robot coverage path planning (mCPP) module, utilizing STC and DARP. This work has been presented, evaluated and tested in the Autonomous Robots journal. The implemented algorithm is optimized in order to efficiently cope with real-life multi-UAV coverage missions, utilizing a novel optimization scheme based on simulated annealing algorithm. The overall methodology achieves state-of-the-art performance in mCPP problem (see comparison with simulated evaluation here).

An on-line instance of an end-to-end mission planner utilizing this mCPP module can be found here: https://choosepath.ddns.net/

In addition, a demonstrative video of the platform can be found here, including examples of missions' creation, management and execution, real-life operations, and indicative results that can be acquired.

The figure above, shows four examples of missions generated by this mCPP algorithm and visualized through the mission planning platform mentioned above.

Highlights:

• Support of multiple robots/vehicles
• Support of convex and non-convex, very complex-shaped polygon regions
• Support of multiple obstacles inside the operational area
• Fair and proportional area allocation of the overall region to each robot/vehicle
• Utilization of the initial positions of the vehicles for both the area allocation and the path planning, in order to make the methodology more efficient
• Energy aware features (paths' length reduction, turns reduction, avoidance of redundant movements that do not contribute in the coverage, etc.)

Input/Output:

The algorithm receives as input the following:

• The number of robots/vehicles
• The desired scanning density (scanning density corresponds to the desired distance between two sequential trajectories in meters)
• A polygon Region of Interest (ROI), formatted in WGS84 coordinate system
• A set of obstacles (polygons formatted in WGS84 coordinate system) inside the ROI (optional)
• A boolean variable named pathsStrictlyInPoly, to select mode between (paths strictly in poly/better coverage)
• The initial positions of the vehicles (optional - if not provided, random will be used instead | Note that the number of the initial positions should always be the same as the number of robots/vehicles)
• The desired percentages for proportional area allocation (optional - if not provided, equal will be used instead | Note that the number of the percentages should always be the same as the number of robots/vehicles and their sum should be 1)

As an output, the algorithm provides set of waypoints (path), for each vehicle involved in the mission, in order to cooperatively completely cover the ROI.

Run the project:

In the src/main/resources folder is included a JSON file, containing input parameters for an example mission with 3 vehicles. The input variables are included in the JSON with the same order as described above.

1st way: run the jar file

In the out/artifacts/mCPP_optimized_DARP_jar you can find the project packed in a jar file. The jar expects as an input the path for a JSON file as the one included in the resources. To run the jar from its current location with the json in the resources folder run:

java -jar mCPP-optimized-DARP.jar "../../../src/main/resources/inputVariables.json"

2nd way: run the Main.java

The Main class of the project expects as argument the path for such a JSON file as well. You can find the JSON file needed to run the main in the resources folder, as shown below:

"src/main/resources/inputVariables.json"

3rd way: run the Poly2Waypoints.java

In the src/test.java folder you can find the Poly2Waypoints class. In this class you can find the definition of some example input variables to run the project, with the same order as described above.

Additional tools

A simple map tool where you can create polygons over a map and copy the coordinates, and a simple matlab code to visualize the output of this project are also included in the "src/main/resources/" folder.

Relevant/useful material:

Platform's demonstrative video

Platform's on-line hosted instance

Article's page

arXiv page

Paper's results repo

DARP original Java implementation repo

DARP recent Python implementation repo

DARP paper

DARP Medium article

Cite as:

@article{Apostolidis_2022,
	doi = {10.1007/s10514-021-10028-3},
	url = {https://doi.org/10.1007%2Fs10514-021-10028-3},
	year = 2022, 
	month = {jan},	
	publisher = {Springer Science and Business Media {LLC}}, 
	author = {Savvas D. Apostolidis and Pavlos Ch. Kapoutsis and Athanasios Ch. Kapoutsis and Elias B. Kosmatopoulos},	
	title = {Cooperative multi-{UAV} coverage mission planning platform for remote sensing applications}, 
	journal = {Autonomous Robots}
}