10 meter Airship

PX4 Autopilot
1

Hi,
i’m new to this world but i started working in a student team whose goal is to make and unmanned 10m autonomous airship to travel a transoceanic route.
I saw that is little to nothing documentation on airships.
Do you think using a Pixhawk 6X-RT whould be enough? ar should i implement it with also a jetson nano?
Also if u have any well made guide that suites my description (or whatever advice you whould give me) is very well accepted.
thanks everyone

1 Like
2

Hey! @Cicciof its an exciting project indeed!

Things to consider:

  • pixhawk 6x-RT is good!
  • Choosing Jetson else an Rpi as companion computer depends on the mission requirement with the cost of battery weight
  • The airship should have a perfect balance, to respond to the gust winds, as they are always vulnerable to it. Unless you found a method to tackle the gust.
  • Would recommend to use differential motor configuration for manuvering in the airship as it is good for steering in wind conditions more reactive then just flaps
    I hope this helps you out in some way. Please let me know if you need further clarifications.

Happy landings!

3

Hi carbon,
first of all i wanna tank you for your kind and usefull answer.

i understand, this friday we will start to do an estimation and a general overwiev of everything.

my idea was to implement some AI (on rpi or jetson) like tensorflow to train with metereologic data, also use some wind sensors to make pix6x adjust the two differential motors (starboard and port) and also the elevator/rudder to tackle the gust.

If you dont mind me bothering you i have some questions about how to get started.
As i saw on the documentation:

  1. install Qgroundcontroll
  2. modify the source code to make a custom one for my airship via vscode??
  3. test on gazebo classic
  4. upload the custom firmware
  5. basic configuration sensor ecc
  6. im honestly lost

I would like to learn whatever you are willing to teach me, obviously if u have time.
Thanks a lot

4

Sorry for the late reply.

  1. install Qgroundcontrol
  1. modify the source code to make a custom one for my airship via vscode??
    -If your airship behaves like a VTOL or a drone(assuming you are taking dual motors for differential actutation), you can use the default PX4 firmware and just:
  • Select an existing airship airframe (SYS_AUTOSTART parameter).
  • Tune PID gains and control parameters for smooth flight.
  • Adjust actuator and propulsion settings in QGroundControl.
  1. test on gazebo classic:
  1. upload the custom firmware
  • We might have to update mixer files and parameters for throttle and yaw manuvers. Unless you are having a :
    -Unique control surfaces or actuators (e.g., no traditional propellers or different yaw control)
    • Advanced flight logic (like auto-docking, precise altitude control, or custom modes).
    • MAVLink modifications (e.g., custom commands for the airship’s behavior).
  1. basic configuration sensor ecc
  • Yes! those are the fundametnals.
  1. im honestly lost
  • Everyone is! until they make thier own way. You started yours now.
5

Perfect, thank you as always.
I was looking foward to order this week a Holybro pixhawx 6x pro + jetson orin board (16gb or 4gb) or a pixhawk 6x + raspberry pi cm4, to start my journey.
Wich one do you think that will suite best my needs?

Also do i need so far the board to get to the third point ? gazebo simulation?
Or can i do everything on a linux based pc and get started testing everything on gazedo before buying the actual board?

If you dont mind me asking and more importatly you dont mind me bothering you i would ask if could contact you if i get any major doubt.

U’re a kind person

6

Hey Cicciof!
No worries at all, happy to help! your hardware choices are solid, but let’s see the pros and cons:

Jetson Orin (16GB or 4GB) vs. Raspberry Pi CM4:

  • Jetson Orin 16GB - ML-based gust prediction, go with Jetson Orin 16GB
    -real-time AI/ML applications
    - hardware acceleration for deep learning models
    - Keep Tabs on the power consumption - please run a bench test
    - For Jetson Orin 4GB has limited RAM for heavy neural networks

  • Raaspberry Pi CM4 - basic control & telemetry, CM4 is cheaper and power-efficient when compared to Jetson Orin
    - basic sensor processing & networking
    - works well for PX4 communication & telemetry

  • Regarding the SITL its not hardware dependent, you can kick start your similuations post setting up the environment setup which is available in Px4 - Youtube.

Feel free to ask anytime! You’re working on a super exciting project, and I’m happy to help however I can. Looking forward to seeing your airship take flight!

7

I’ll definitely keep you posted.

By the way i have another question: speaking of calibration i saw on the documentation that the only way is to rotate the model on every axis…
This is obviusly impossible with our 10m airship.
Is there another way to do it? (like an airborne calibration)
thnks as always

8

Hey @Cicciof, I no, I haven’t tried airborne calibration, I would recommend to gather more hands, to perform the calibration as the process will be carried out only once in the life cycle of the system.(If you nail it in the first shot.)