Wrapping up more results from my thesis. In this video I show how with a single (guidance) control law a team of drones can intercept an independent target. The team is able to estimate the unknown velocity of the target and to orbit around it with a prescribed shape. The main idea behind is about exploiting (controlled) discrepancies to the distances that the drones want to maintain between each other.
The algorithm sets the accelerations to be tracked by the drone. This is accomplished by just controlling the corresponding attitude angles. The only information needed by the algorithm is just the relative positions (in arbitrary frame of coordinates) between neighboring drones, i.e. there are not relative or absolute velocities involved.
I still have to edit some videos more, next time I will choose happier music :P.
Comments
Hi Zi,
you will find it here: https://github.com/noether . Probably in two or three weekends the code will be there.
I was using the vehicle from here https://vimeo.com/34364994 , but all controls etc are mine (pretty simple and standard ones I would say). You can find them in the last chapter of my thesis (the link is in the blogpost)
Hi Hector,
This is very awesome! What is your github name I will start following you right away.
Did you figure out how to simulate quadcopter on x-plane yourself or is there a tutorial I can follow? If there is a tutorial could you please point me to the right website/community?
Cheers
yep, it is xplane 9 and you can simulate as many quadrotors as you wish. For sure I will upload my code to github, so you can take a look at it. Most probably I will post it here.
Cheers
Hi Hector,
Very interesting work you have done. I am doing something similar for my thesis but I am using PX4 for my autopilot. What simulation are you using in that video? It looks like x-plane but I don't know you can simulate quadcopter in x-plane.
Hi Greer,
unfortunately I believe I will not have time myself for developing to others autopilots. Nevertheless you will see all these algorithms (soon) in Paparazzi as an "independent" modules. I do not know whether you are familiar with Paparazzi, but by a module we mean that the algorithm source code is quite independent on the platform. Somebody familiar with the APM stack (unfortunately I am not) should be able to port them quite straightforwardly.
And of course, whatever question regarding about how the algorithm works, or how it should be implemented, can be asked in the gitter chat of Paparazzi or in our mail list, I will be happy to help :P.
Hi Hector - I see a lot of your research has been using the paparazzi autopilot. Do you have any plans to develop this to other autopilots like APM or to other C2 protocols like mavlink?