9 axis IMU

[zik]

Here's a side project I've been working on. It's an IMU with 3 accelerometer axes, 3 gyro axes and 3 magnetometer axes. It'll have an on-board processor to do kalman filtering and it'll communicate using either serial, I2C or SPI. The user can choose how they want to communicate with it.

I figure this will be useful not just for my Flying Fox autopilot, but it should hook up to paparazzi or any other autopilot pretty easily. I've designed it to be nice and small - so it should be able to fit into small airframes. How small is small? Think less space than a MicroMag3 (but it'll do a lot more!)

Currently this design is in the early stages. If there's enough interest I may continue with the design and go ahead and produce it for other people to use. Let me know if you're interested.

I haven't decided whether to open source this one or not yet.

[small_rcer]

Hello Zik;

or should that be

G'day Zik;

Depending on how good the code is, and the cpu power employed, this could be very useful to many projects. I see it being applicable to many types of autonomous vehicles. Think of Underwater where GPS is unusable, or in buildings where GPS signals are too intermittent to be useful. As we all think airplanes here, it is a given that they are useful in our play pen. If it can connect to Atto, Papparazzi, your own Flying Fox and others it could be most useful.

I am torn wether it should be open source or not. As you have just published a potential schematic, the cat is sort of out of the bag as it is, but at the same time you should be rewarded for your efforts. The upside is, if the software is open source, there are people who could accelerate the completion and thus speed up the adoption.

If the hardware is complex and dense enough to preclude all but the most dedicated DIY person, then the code could be free as in Asterisk, and the hardware could be proprietary as in Digium and Sonoma voip telephony products.

People would be free to change the code and modify the behaviour, but you design and license the hardware to other manufacturers or resellers.

Either way bring it on. It is likely going to be needed sooner rather than later.

Jim H

[Clarkst]

Very intersting. I say Go Go Go!! Put me on your list of early adopters/purchaser/beta tester. I2C would be great.

[clolson]

My two cents ...

There are a number of devices that provide 3 axis gyro, 3 axis accelerometer, and a few that have 3 (or 2 axis) magnetometers. If you can provide excellent quality sensor fusion code that can achieve results that are comparable to something like a Microbotics MIDG-II then I think you'll really have something that stands out. However, based on my limited understanding, the MIDG is able to achieve it's superior results because it also incorporates gps based direction of movement data. What you are proposing might be popular if you can beat the prices of the available devices by a good margin (and have the same quality and robustness.)

What I find when I look at available hardware is that you can buy all the pieces, but every component requires some sort of serial port ... the IMU, then gps, then servo control, communication with flight computer, communication with ground station, etc. If I spec out something that is built from pieces, I find myself needing at least 4 serial ports, and many of the available embedded computers do not provide that many.

So a device that I would consider most valuable for UAV developers would be something similar to the Xbow MNAV (now discontinuned) that provides IMU + GPS + servo control + air data + some cpu/programability all in one tight package.

The O-Navi Phoenix is the only other device I've been able to find that seems to tie all these things together, but that comes with *very* limited example code as I understand.

The holy grail of UAV's as I see it would be something that pulls together accelerometers, gyros, magnetometers, gps, air data, and servo control into a tightly integrated package, and comes with a quality algorithm to fuse all the sensor data together into an accurate location and attitude estimate (especially while performing arbitrary flight maneuvers.)

The MIDG-II is about $6750 last time I heard the price, and that only provides IMU + GPS. No air data, no servo control. The Xbow MNAV was about $1700 and came with some code, but the code had severe limitations and was nothing to get excited about. The other complaint I have with the MNAV is that it has some robustness and reliability issues. If you could put an IMU based package together that does everything for < $1000, I bet you would have a real winner.

Curt.

[zik]

The holy grail of UAV's as I see it would be something that pulls together accelerometers, gyros, magnetometers, gps, air data, and servo control into a tightly integrated package...
Thanks Curt. Sounds like Flying Fox plus this IMU could provide exactly what you need. Check back in a few months and you might be in luck.

[zik]

Depending on how good the code is, and the cpu power employed, this could be very useful to many projects.
On the CPU power - the dsPIC processor should be perfect for Kalman filtering computation. But keep in mind that this is just a schematic I threw together yesterday (a sick day off work) so the chances are it'll need a lot of testing and refinement before it's finished.

I am torn whether it should be open source or not.
Yeah I'm torn too actually. I might see how this one goes as a commercial project to see if there's any money to be made doing this kind of thing. If I can make a living making this kind of stuff then I'll be able to put a lot more of these kinds of devices out.

Cheers,
Zik