We've been dropping hints about the forthcoming Linux-based versions of APM/Pixhawk. Here's another glimpse of the future, using the forthcoming Pixhawk Fire "cape" (like an Arduino shield) for Beaglebone Black. Victor Mayoral Vilches explains the BeaglePilot project, which is a Linux-based autopilot for flying robots. It was a project selected by BeagleBoard at the Google Summer of Code 2014.
BeaglePilot project aims to create the first Linux-based autopilot for flying robots using the BeagleBone and the BeagleBone Black as the “hardware blueprint”. For this purpose the project will focus on integrating ArduPilot (most popular autopilot) in the BeagleBone (Black). The main tasks performed will be userspace drivers development, the Robot Operative System Integration (ROS), web IDE exploration and security assessment. The BeagleBone cape “PixHawk Fire Cape” and the robot “Erle” will be used.
Comments
Pescadito01 you posted the same question in three different threads. Man, are you kidding?
Besides this matter, i'd say it's a pretty interesting platform. I've seen SparkFun has some "capes" already available that includes one of the IMUs that we plan to support. I'd be happy to give it a push on the development side if someone can get his hands on some boards.
If they release the schematics and the layout it could become a strong option for future hardware autopilots.
Please, could anyone give a opinion if the new Intel Edison could be a good platform to make a linux autopilot board like the newsly development of BeaglePilot by diydrones and others?
Following this with anxious anticipation. I honestly wish I could help, but I have ZERO knowledge (yet) of BBB.
Hi everyone,
@Chris thanks for sharing the presentation. I'm really amazed to see how much attention it has acquired in a few hours. I'll try to address some of the questions and comments:
Gary, you're right is quite aggressive but we will do our best :).
@Mustafa, I use Sublime Text.
@Jack, 3DRobotics is doing it. Anuj and Siddharth are being funded by 3DR.
Anuj and Siddharth will be working on the PRU side. They will create a layer of abstraction for the PRU using the TI StarterWare code. Please refer to https://github.com/BeaglePilot/beaglepilot for details.
As @hotelzululima pointed out, the design PXF is already available (still under test). I re-designed the robot Erle to include the same sensors as the PXF. I'm working with both boards.
Anyone interested is welcome to join our community. We strongly believe that this port to Linux will become one the key blueprints for the future of robotics :)!
just looked at the readme at https://github.com/diydrones/PXF because I was going to submit the design to tindie.com for a bespoke build to et some hardware or even to oshpark to at least get boards I can populate myself says not to use for new boards and that a private repo holds the current designs for same...
sigh
hotel zulu lima
ps not so open yet
@Guy I had a brief look at the PRUs a while back - they seemed fairly poorly documented and difficult to use, but if that has changed, they may be a viable alternative.
When will the cape be available?
From a quick glance at the cape design, it looks like you're connecting the sensors directly to the BBB communication peripherals (I2C,SPI,UART etc) i.e. without a separate micro dedicated to low level communications with the sensors - is this the case? If so, you'll have to play with real-time patches to ensure hard real-time sensor sampling which often requires strict timing constraints. Further, I imagine a lot of cycles will be being wasted, constantly interrupting out of the main code to sample the sensors.
In my opinion, rather than taking steps to make the BBB more like a low level micro, it's better to actually add a micro to the cape for at least sensor sampling then use the BBB for what it's good for (high level processing without strict timing constraints + all the other luxuries of Linux). You could add servo PWM generation to the micro and even a simple failsafe autopilot in the event that Linux crashes or hangs.
My own experience with a BBB-based autopilot is that low latency serial comms uses more CPU load than you'd think.
The BBB is fantastic, good luck!
Why doesn't 3D Robotics fund it instead of making them go through the Goog summer of code?
I Join the group of guys that would fly with linux based apm :)
-
1
-
2
of 2 Next