I just got back from Paris, where I was meeting with the Parrot team, who are now part of the DroneCode project. Here's the first glimpse of what they are working on: the APM code running on the Bebop drone! The picture was taken at Parrot's HQ in Paris; the team that did the work is in the background.
Right now it's in an early alpha (here), but it was flying very well. That's me (with hat) on the Logitech controller in the background. Any twitchy moves are all pilot error ;-)
Comments
Awesome news !
One of the benefit of Bebop is that the bottom camera act as flow sensor for non GPS environment.
Hope those Linux multicopters will widely open to flow sensor and other instruments for reliable non GPS operation.
Chris you look like you're on your way to Wembley to watch Wayne Rooney.
Yes, the Bebop port only took JulienB 24 commits! I think this highlights a bunch of things including JulienB's effort/ability and also how *relatively* easy it is to port APM to a new board. That work was of course built upon the ground breaking work of ErleRobotics and emlid.com's NAVIO board.
Great stuff, really really happy to see the AP_HAL_Linux working in the Bebop :)!. Congrats the Parrot team, Tridge and Randy for getting this working.
Just to show some repeatability here's the Bebop flying in Karuizawa. This is with MissionPlanner + Joystick, Copter3.4-dev with the patches from Julien at Parrot (and a couple from Tridge too). We haven't successfully got GPS based modes working (Loiter, Auto, etc) but it won't be long probalby.
I like to see post like these! Good to see different realities working together on a common project: the private company that allow a open source software on their (good) hardware. It will be wonderful if they add mavlink and waypoint capabilities to the bepop! :)
For the wifi is adequate fot the bepop? I've tried the bepop and i lost the wifi connection for 4 minutes good (in that time the bepop stay on position and land automatically due the battery at the 4th minute). The owner says it was the first time that the ipad app / skycontroller lost the connection. Wifi for me is not good as primary link, due to the protocol is too risky is certain situation.
have a nice day!
Randy and I are working a lot with Julien from Parrot to get the Bebop port merged. The current stumbling block is an issue where the EKF gyro bias grows too large in some flights causing attitude instability. Once we solve that we expect to merge it as a port that anyone can use.
It has been great working with the Parrot developers on the port, and I really look forward to having the Bebop as a platform that anyone can use ArduPilot on.
It is a great little copter, and may help to get the Linux port of ArduPilot more exposure. In the future we'd like to get the video system working and support the builtin sonar and optical flow sensors as well.
This port, along with the PXF, NavIO and BBBMini ports means users have a lot of choice if they want to experiment with Linux based vehicles. Many thanks to Parrot for taking the initiative with this port!
Cheers, Tridge
Gary and Chris... thank you both for the input. Agreed on APM opening new methods for control. One thing that the original AR Drone had going was a modding/hacking community, the likes of which I have not seen yet with the Bebop. But, its still young.I really like the idea of the Bebop to throw in a backpack and go! As Chris mentions, this is where it shines. I may wait until the mk2 as well.
Thank You Chris,
I have been paying close attention to the BeBop and I know the Achilles heel is getting the digital stabilization to work in roll.
Pitch and Yaw are easy - on an oversize camera chip like they are using they are simply and X or Y pixel offset.
Roll however is axial, and not even on a tightly defined center of rotation.
And combining roll with yaw and pitch an even bigger nightmare - I'll bet if you talk with there devs they will be nodding their heads.
Computationally a whole different ball game.
I think a lot of the reason in the delay in coming out with it was getting that to work acceptably (fast enough computationally).
My Sony AS100V does X and Y stabilization just great - Axial not at all.
I am very happy to hear that they have cracked this serious nut as well as they have.
Also interesting to hear your take on the Sky Controller, it does seem ponderous.
Till your response I considered the Sky Controller to be pretty mandatory, but now not at all.
I will probably wait till BeBop 2 to see what they add, my guess is they will pass into the bounds of a truly capable tiny and safe platform.
In the meantime I am working on rolling some of my own.
And HeliStorm, for now I think WiFi is probably adequate for BeBop (given realistic limitations as represented by Chris.)
If you want to jack up the range there are inexpensive (and small) WiFi amps and antennas available.
I would prefer a standard radio link too (albeit I value the real time video preview as well).
Once they have interfaced APM, if they make it available, my guess is a bit of software and hardware hacking could make a RC receiver link work through APM regardless of what Parrot intended.
It definitely looks like the best small thing out there at this time and at $500.00 for the non-SkyController version it's a lot cheaper than any of the bigger, less safe quadcopters out there.
And Jack,
One thing you can say about Parrot is that they seem pretty determined to lead rather than just follow along and my guess is that their firmware has no particular vested interest to them.
If APM can work out better they will do it, getting digital camera stabilization to work at all well on one of these things is a major win, even Canon and Sony haven't managed it - yet.
Best Regards,
Gary
-
1
-
2
of 2 Next