Survey of open-source autopilot systems

I would like your feedback on the following analysis.
Among the huge number of projects and forks, it is extremely difficult to get a clear idea of the activity and evolution of autopilot systems for UAV.
For this purpose I made a survey of the main open-source autopilots based on an automatic analysis of their source code repository (more info in my phd thesis here, p. 23).
For example, here is the evolution of the number of contributors:

… and the number of commit per week (mean filter with 6 months sliding window):

On the figures we can see the last update date of the projects (thus if they are discontinued or not), the general trends, the forks (see e.g. how the cleanFlight fork of Raceflight has evolved, itself a fork of baseflight…), etc.

The questions are:

  • Does that spark your interest?
  • Would you like to see a website were one can see these figures automatically updated, compare projects, add code repositories, etc. ?
  • If yes what would be your suggestions?


If there are some SWAT analysis or comparison that would be great.

And most importantly, in English :slight_smile:

It is nice to see what projects are active! I think that once you are in the top 5 or 6, the intended use will be the main factor. For research, the PX4 platform has a better integration with motion capture and ROS for example.

Yes, the graphs does not show anything about the features or strengths of autopilots, only stats. And it is not a ranking of autopilots. But indeed, it highlights active projects or shows, e.g., that PX4 contributors are apparently more active than Ardupilot ones, as the number of commit/week is similar with fewer contributors (at least up to 2017).

It’s interesting to see studies like this. Thanks for sharing!
A technical comparison would be useful though.

It would be really interesting to see how the number of contributions changes with the release of new hardware, I know a lot of people avoid ardupilot because its hardware counterparts aren’t powerful enough.