Multi UAV test area coverage

A multi UAV test platform for area coverage path planning algorithms.

Introduction

The multi UAV test area coverage repository is a set of tools that is part of an experimental platform aimed at evaluating different replanning algorithms for multi-agent flight systems.

It consists of the following components:

• planner: plans a mission given a region to inspect and the number of drones to use.

• execution_monitor: drone monitors that detect important events such as the drone getting lost or finishing the inspection.

• performance_evaluator: generates useful metrics such as the total time of the mission or the total distance covered.

• performance_viewer: graphic interface that shows the current state of the mission and its metrics.

• multi_robot_simulator: simulates the mission in a photorealistic environment using Flightmare. It also simulates any unexpected events that can occur to a drone. The events currently supported are: BLIND_CAMERA, GO_HOMEBASE, BATTERY_DISCHARGE, FALL_DOWN, STOPPED, RECTILINEAR_MOTION, SLOW_MOTION.

Installation

Install Flightmare

The multi_robot_simulator uses Flightmare to simulate the mission. To use it, you need to install Flightmare with ROS.

After installing it, if you want to use the solar farm scene used in the examples, you need to install the modified Flightmare binary from here (link), extract it into the /path/to/flightmare/flightrender and modify the following file:

• flightmare/flightlib/include/flightlib/bridges/unity_message_types.hpp

// In UnityScene Enum add the ID for the solar farm scene and increase the number of scenes

enum UnityScene {
  INDUSTRIAL = 0,
  WAREHOUSE = 1,
  GARAGE = 2,
  TUNELS = 4,
  NATUREFOREST = 3,
  SOLARFARM = 5,
  // total number of environment
  SceneNum = 6
};

To use multiple quadrotors in Flightmare you also need to change the following files:

• rpg_quadrotor_common/quadrotor_common/include/quadrotor_common/trajectory_point.h

// Add the definition of the copy constructor
TrajectoryPoint(const TrajectoryPoint& other);

• rpg_quadrotor_common/quadrotor_common/src/trajectory_point.cpp

// Implement the copy constructor

TrajectoryPoint::TrajectoryPoint(const TrajectoryPoint& other)
: time_from_start(other.time_from_start)
, position(other.position)
, orientation(other.orientation)
, velocity(other.velocity)
, acceleration(other.acceleration)
, jerk(other.jerk)
, snap(other.snap)
, bodyrates(other.bodyrates)
, angular_acceleration(other.angular_acceleration)
, angular_jerk(other.angular_jerk)
, angular_snap(other.angular_snap)
, heading(other.heading)
, heading_rate(other.heading_rate)
, heading_acceleration(other.heading_acceleration)
{}

• rpg_quadrotor_control/trajectory_planning/polynomial_trajectories/include/polynomial_trajectories/polynomial_trajectory.h

// Add the definition of the copy constructor
PolynomialTrajectory(const PolynomialTrajectory& other);

• rpg_quadrotor_control/trajectory_planning/polynomial_trajectories/src/polynomial_trajectory.cpp

// Implement the copy constructor

PolynomialTrajectory::PolynomialTrajectory(const PolynomialTrajectory& other)
: trajectory_type(other.trajectory_type)
, coeff(other.coeff)
, T(other.T)
, start_state(other.start_state)
, end_state(other.end_state)
, number_of_segments(other.number_of_segments)
, segment_times(other.segment_times)
, optimization_cost(other.optimization_cost)
{}

Install dependencies

Install the PySide6 library with the following command if you want to use the performance_viewer, and tmux to execute the examples:

pip install pyside6
sudo apt install tmux

MUTAC workspace

Create a catkin workspace with:

cd
mkdir -p catkin_ws/src

Now, inside the workspace you can clone the MUTAC repository.

cd ~/catkin_ws/src
git clone https://github.com/cvar-upm/multi_uav_test_area_coverage .

Build the workspace:

catkin build

To finish, add the workspace setup to the .bashrc file and create the MUTAC_WS environment variable for the workspace.

echo "source ~/catkin_ws/devel/setup.bash" >> ~/.bashrc
echo "export MUTAC_WS=~/catkin_ws/src" >> ~/.bashrc
source ~/.bashrc

After this installation you can now execute every example in the test package.

Usage

Running Tests

The framework exposes several mission cases for testing multiarea coverage planners. These cases are located in the tests/n_drones/casei directory, n the size of the tested swarm and i the test case.

Each test is composed of a config.yaml and examples_events.csv files defining the mission and a mission_launch.sh launch script. Executing the mission_launch.sh script initializes the test. For instance, the test case 1 for 3 agents invocation would look like:

$ ./tests/3_drones/case1/mission_launch.sh

On mission initialization, a tmux session is created with different tabs for the planner, multi_robot_simulator, performance_evaluation, performance_viewer, global_monitor and execution_monitors.

The performance_viewer instance shows a graphic interface with mission data and metrics while the multi_robot_simulator renders the simulation scenario and the camera view of each agent.

Mission will not start until the planner's generate_plan service is invoked with the input region. This starts the system with the config.yaml parameters.

Once the mission is finished, the tmux session can be killed with Ctrl + B and typing :kill-sesssion in the terminal.

Modifying Tests

New tests can be created by modifying existing ones. The config.yaml file inside a test case contains the starting swarm, the location of the events file and launch parameters of the planner. An events file holds all the unexpected events that will happen during the mission. By default it is the example_events.csv next to the config.yaml file in each test case.

The availiable entries in the config.yaml are:

Entry	Type	Description	Example
n_drones	Integer	Size of the initial swarm	3
homebase/dronei	List	Home of agent i	[590, 90, 325]
mission/height	Float	Inspection Height	370
mission/distance	Float	Path Separation	10
common_space/height	Float	Positioning Base Height	375
common_space/height_increment	Float	Positioning Safety Distance	5
update_plan/threshold	Float	Replanning Threshold	0.8
events/file_path	String	Events File	/tmp/example_events.csv

The events file contains a sheet with all the events that may happen during execution. Each event declaration specifies the target agent, the firing instant and the event itself with its corresponding parameters. A discharge of a 5% of the agent with identifier 1 after 15 seconds of execution would be represented as:

1,15,BATTERY_DISCHARGE,5

The availiable events are:

Event	Parameter	Description
BATTERY_DISCHARGE	Percentage	Reduces the batery of the agent by the percentage.
FALL_DOWN	None	Forces the agent to fall down.
RECTILINEAR_MOTION	None	Forces the agent to move in a line.
STOPPED	Time	Forces the agent to stop for the specified time.
GO_HOMEBASE	None	Forces the agent to return home.
BLIND_CAMERA	None	Deactivates the agent camera.
SLOW_MOTION	None	Reduces the agent movement speed.