DRONE EXPERIMENT

This is a Genom3 based experiment for drones in search and survey application. The experiment depends on Genom3 framework, Robotpkg modules. The experiment is using Unified Planning for planning and OpenCV for computer vision.

Pre-requesties

• Ubuntu >= 20.04
• Gazebo 11
• Python 3.8
• Pip >= 22.3
• OpenCV4
• Docker (Optional)

The experiment has been tested on Python3.8 / 3.10, Ubuntu 20.04 / 22.04 and Gazebo 11 with the system architecture x86_64/AMD64.

Setup

Using Docker

One can use the docker image to setup the experiment. The docker image is available in Docker Hub. To setup the experiment using docker, we can use the following commands:

# Build the image
docker build -t genom3-experiment .

# Run the image
./start.sh

The docker container is a VNC server. So, we can use any VNC client to connect to the docker container. The address and port of the VNC server is localhost:6080. The password for the VNC server is ubuntu.

Note: Inside the docker container, it is expected to install the python dependencies everytime. The python dependencies are not installed in the docker image due to active development.

From Source

To setup the experiment, we should already have ROS and Gazebo installed. If not, please follow the installation instructions for ROS and Gazebo. Once ROS and Gazebo are installed and configured, we can install the experiment.

To setup the experiment workspace, we can use the following commands:

# Setup the workspace
mkdir -p drone-experiment && cd drone-experiment
git clone https://github.com/lerema/genom3-experiment.git

The easier way to install genom3 and robotpkg modules is to launch the following commands:

# Move to the root of the repository
cd genom3-experiment

# Setup the experiment in Ubuntu 20.04
bash install-scripts/setup-experiment.sh --all

# Setup the experiment in Ubuntu 22.04
bash install-scripts/setup-experiment-source.sh --all

Note: The setup script asks for su permission to install robotpkg and ros packages using system package manager

The setup-experiment.sh script has the following options:

Usage: bash install-scripts/setup-experiment.sh [--help] [--clean] [--genom] [--robotpkg] [--all] [--update]
  --help: show this help
  --clean: clean the experiment
  --genom: install/reinstall the genom3 modules
  --robotpkg: install/reinstall the robotpkg modules
  --all: install/reinstall all the modules
  --update: update the experiment

Note: setup-experiment.sh --clean will remove the installed modules related to the experiment. But the environment variables in ~/.bashrc will not be removed. So, we need to remove the environment variables manually.

Python

The experiment depends on Python3 by default. To install the python api, we can use the following command:

# In the root of the repository
# Upgrade pip to the latest version. One of the dependency uses `pyproject.toml` which is supported by pip >= 22.3
python3 -m pip install --upgrade pip

# Install the drone python api
python3 -m pip install .

Note: Once the setup is finished, we can relaunch the terminal or run source ~/.bashrc to update the environment variables.

Usage

To start the experiment, you need to launch the files from run/ directory. For example, the following command starts the experiment with a single drone:

# To start the world with single drone equipped with camera and AR marker based environment
drone_experiment && bash run/start-simu-ar.sh --python # --python/--tcl. --python is optional

The above command will load the gazebo world, rviz and genom3 components. The terminal will be launching each component along with a empty terminal for the user. Please resize the terminal to your comfort. The user can use the empty terminal to run the python api or the UP demo.

To start the sample experiment with the python api, run the following command:

# Simple example for a single robotdrone
# From the root of the repository
python3 src/sample_app.py

To start the UP demo, you could run the following command:

# UP Demo script
# From the root of the repository
python3 src/up_demo.py

Note: To restart the experiment, it is recommended to close all the terminals and relaunch the experiment.

Current Status

Currently, the experiment is in the development phase. The following are the current status of the experiment:

• Python API for a single drone
• UP demo for a single drone
• Python API for multiple drones
• UP demo for multiple drones
• Improve planning problem
• Implement Hierarchical Planning