Multi Vehicle Simulation Envirnoment

The goal of the following tools collection is to allow for the MAVLink communication to be done over MQTT instead of UDP or TCP, thus making over the internet drone management easier. The implementation of the tools occours in the form of external script to be run aside of the specific MAVLink communicating instance (PX4, QGC, API, ecc)

Original Communication Scheme

MQTT based communication scheme

Working Principle

MAVLink Connection

MAVLink connection communicates by sending packets through UDP. The default receiving ports are 14550 for the ground control station and 18570 for the PX4-Autopilot. In case of multiple devices, at the QGC side, tha same receiving socket is used for all the incoming data, while the data is sent to the different instances through UDP ports 1857x (with x the instance number). In case of more than 10 devices, the UDP port 18579 is used for all the excedence.
In order for a communication to be established between the ground control station and the device, a heartbeat stream must be sent by both sides.

MAVLink Connection with PX4-Autopliots

A socket is created at the QGC side, listening for port 1857x (x the sim N) and sending on 14550 to QGC. The received message is published with the topic /(api-key)/0/qgc_to_px4_x, while mqtt messages received with topic /(api-key)/x/px4_to_qgc are sent over UDP 14550 to QGC. At the PX4 side, since all the PX4 instances are supposed to run on different devices, the mqtt received messages from QGC to PX4 are sent via UDP to port 18570, from the same socket that is listening on 14550 for PX4 to QGC packets. This way there is no need of modifying the default ports in PX4 simulators.

MAVLink Connection with External Devices

If in the QGC configuration file ENABLE_EDEV is set to True, messages with topic /(api-key)/10x/edev_to_qgc are received, the information contained in the plain text payload is decoded and saved.
The payload structure is lat|xx.yyy|lon|xx.yy|ele|xx.yyy|h|xx.yyy|v|xx.yyy|m|xx.yyy

A pymavlink drone instance is created for every edev (to be set in the EDEV_N param), and every time an MQTT message is received, a GLOBAL_POSITION_INT Mavlink message is generated, along with a Heartbeat. External devices are seen from QGC as FREE BALLOONS, so no errors will be displayed because of the unilateral communication.

Communication to Fiware

By setting ENABLE_FIWARE in px4_conf.json to True, the messages received from the PX4-Autopilot are decoded. When the MAVLink Message ID is 33, meaning GLOBAL_POSITION_INT, the message is decoded and the extracted values are published under the topic /(api-key)/x/attr with payload structure lat|xx.yyy|lon|xx.yy|ele|xx.yyy|h|xx.yyy|v|xx.yyy|m|xx.yyy

API

If ENABLE_API is set to True in the PX4 config file, a communication is instanciated over MQTT to and from the ports 14540 and 14580 at the PX4 side, allowing to control in real time the mission from apis. Live update of the fences is possible, and doesn't require mission to be halted.

Envirnoment Set-Up

Clone MVSE Tools

git clone https://github.com/FlexiGroBots-H2020/MVSE.git

PX4 + jMAVSim setup

Requirements: Ubuntu 20.xx LTS, git, pip, paho-mqtt

• Clone PX4-Autopilot suite and run the setup tool

  git clone https://github.com/PX4/PX4-Autopilot.git --recursive
  bash ./PX4-Autopilot/Tools/setup/ubuntu.sh
  

• Copy content of /MVSE/PX4 in /PX4-Autopilot/MQTT

• Edit empty_px4_conf.json to match instance value and MQTT address and credential and rename it px4_conf.json

• Open a terminal in /PX4-Autopilot/MQTT and run px4_mqtt_client.py
  Make sure to choose a free instance number

  python3 px4_mqtt_client.py
  

The script will edit the config files in the PX4 source and start the MQTT client, close it if you don't need to run the simulation.

QGroundControl fist setup

Libraries requirements (python):paho-mqtt, pymavlink
(tested with Ubuntu 20.04 LTS, for different distributions commands may differ)

Ubuntu

• In a terminal execute

  sudo usermod –a –G dialout $USER
  sudo apt-get remove modemmanger –y
  sudo apt install gstreamer1.0-plugins-bad gstreamer1.0-libav gstreamer1.0-gl –y
  

• In folder /MVSE/QGC/ right-click on QGroundControl.AppImage -> Permissions -> Allow execute as a program
• In empty_qgc_conf.json, change values of DEV_N and QGC_INST accordingly to environment and rename to qgc_conf.json

Windows

• Dowload and install QGroundControl from here
• In MVSE/QGC/empty_qgc_conf.json, change values of DEV_N and QGC_INST, fill MQTT broker data accordingly to environment and rename to qgc_conf.json

Simulation

PX4 + jMAVSim Simulation

Open a terminal PX4-Autopilot and PX4-Autopilot/MQTT

Start PX4-Autopilot and jMAVSim from /PX4-Autopilot/ terminal

make px4_sitl_default jmavsim

Start PX4 MQTT Client from /PX4-Autopilot/MQTT terminal

python3 /MQTT/px4_mqtt_client.py

QGroundControl

Note: If multiple instances of QGC are running, make sure that only one has Emit Heartbeat checked in QGC -> Settings -> MAVLink

• Run QGroundControl
• Open a terminal in /MVSE/QGC
  Start QGC MQTT Client

  python3 qgc_mqtt_client.py