Crash 4 Sure: integrated Autopilot & OSD



A few months earlier, mmormota and I have started to develop a combination of autopilot and OSD. We're made it for FPV purposes mainly, but you may have found it interesting because of the autopilot part. We've just accomplished our first flight tests in this week and the system didn't served it's temporary name: Crash 4 Sure :)
On the picture above you can see as it's going round on a small rectangle course. Here is the video: http://vimeo.com/15128879

Our main goals are:
- Simple installation, we hate the mess of wires.
- Keeping the part count and prices low.
- Make it modular for further expansions, but the basic unit must able to stabilize and bring back the plane safely.

Our unit is most likely a single chip design, based on an STM32 (ARM Cortex M3) controller witch performs all the tasks. This combination of the two systems is quite convenient. For example we can see clearly how the autopilot works, is the orientation correct, because it can display everything on screen real-time.
The sensing of orientation is IMU based, I'm currently using the Razor 6DOF IMU board from Sparkfun and mmormota is using ITG-3200 and BMA-180 breakouts for this task.

The autopilot have the following modes of operation:
-Manual mode
-Stabilized mode (exactly the same as ardupilot's "fly by wire")
-Waypoint mode
-Return to Home

We want to expand it's abilities with automatic takeoff and landing, but first we want to make the current software and features as safe as possible. We were using Raisonance's Ride7 and Rlink for he development, but run into the 32k code limit and have to transfer the project under Eclipse + GDB + OpenOCD + Rlink. Programming and debugging the target still didn't work very well, this environment is terribly complex and hard to configure everything right...
The entire code is in C, using floating point arithmetic everywhere. Now it takes about 10% CPU time to run everything what's related to the autopilot and the OSD consumes about 15% CPU time, so there is still plenty of possibility in the hardware. Ohh, and the core runs on 48MHz instead of the maximal 72MHz.

So this is the whole story in a nutshell, we're want to thank all of YOU who maintain this great community.

Special thanks to William Premerlani for the great papers !

Here are some photos of my equipment:

p1010077ml.jpg


p1010079f.jpg


p1010132b.jpg


p1010122b.jpg


E-mail me when people leave their comments –

You need to be a member of diydrones to add comments!

Join diydrones

Comments

  • In this case an open source, software based OSD would be much reasonable.
    For exaple: http://www.rcgroups.com/forums/showthread.php?t=1290963
  • +10 for James comment
    I think James is interested by the OSD part.
  • Try Remzibi concept - open shematics, closed source
    We're want to do it exactly the same way. FYI it's hex only AND locked, so you can't program an other controller with the provided hex file, only the ones which you buy from Remzibi. The point is, you can't made a Remzibi OSD form the given schematics and firmware.

    If you could integrate it with APM
    Integrate it with APM ?!
    That doesn't make sense. C4S is all ready an IMU based autopilot, why do you want to connect it to APM?
  • If you could integrate it with APM and make it open source, you could probably get Chris to sell it here in the DIY store.
  • Sys-Ak, don't wait until it is good enough - start to sell it. Try Remzibi concept - open shematics, closed source (hex only or locked chips) and probably closed configuration software with automatic updates, bootloader, etc. Also good online manual, examples and videos and keep discussion alive. Sell it via forums, etc. using Paypal and I guess you will be surprised from results, in a good sense :-)
    My two cents, as some here used to say...
  • For now, it's only a hobby project. We're working on it and will see what happen.

    If it'll be good enough, we would sell it.
    If it won't be so good, competitive, if we don't have enough time or simply loosing interest, then we will publish everything.

    So it's hard to say in this community, but it will remain a closed source project.

    If anyone had a good idea, how to made it open source and earn some money with it at the same time, don't forget to post it!
  • very nice. I would love to see the code and try this out
  • Very nice and clean motherboard concept! Hope more details like schematics and code will be shared if it is not intended to be a commercial project. Video generation without special ASIC like Maxim one is definitely interesting. Keep the good work!
  • Is video getting on and off the board through the ribbon cable?
    To be more precise, there is an other 2x3 pin connector at the bottom of the board, facing forward. The board is secured by two C shaped guides glued in the canopy, so the whole thing can be slide in and connect to the camera and GPS automatically. Unfortunately this idea come in to my mind when I was assembling the canopy, so it was made afterwards and don't look so good. I'm including a picture of that.

    I like where you went with the "motherboard" concept.
    That was made to achieve some quick success and if somethings wrong, it can be modified easily. As mmormota said, we will made a small single board, but I wanted to fly with it earlier. Here are some pictures of his motherboard too. ;)

    probapanel5.jpg


    probapanel1.jpg


    p1010135wj.jpg

  • Thank you all for the kind words.
    Sys-Ak is sleeping now - I try to answer the questions.

    Are you doing video sync generation and all of that stuff with software?
    Yes. A DMA channel feeds the SPI port so the core loading is minimal.

    Is video getting on and off the board through the ribbon cable?
    Yes.

    What are you using to inject the OSD into the composit video?
    Just a single resistor. :-) Maybe later we choose a bit more sophisticated solution...

    I like where you went with the "motherboard" concept.
    Later we are about to make a single board version.

    Would have been nice to see some of the activity during the "months earlier"
    I uploaded a video several weeks ago (check my videos) - that time only the stabilized mode was ready, no waypoints etc.
    Before that time not too much happened - coding, debugging etc.
    We plan much more test flights, hopefully better quality videos as well. After several succesful sessions -as we can trust more in the code - we risk a better camera, right now the cheapest one works onboard... :-)
This reply was deleted.