new autonav / GPS / IMU / OSD !!!

[iflybyu77]

Just so you know, we ARE reading and paying attention, so don't think you are wasting your time. You seem pretty excited about the development, so I'll cheer you on! Seems like a pretty neat idea going modular..

[xStatiCa]

yea... what he said :). Very interested in something that is open to develop on myself. Keep the info coming.

[Hogster]

So basically this would be a drop-in replacement for something like the Helicommand? How about flybarless mixing?

I love the idea of GPS hold! Keep up the great work with the development :)


David

[corvette321]

well helicommand is approx 600 smackers!!! ouch...

this is an OSD for 100-125 modular also.. that you can interface anything too :) and develop more past what we have done in the firmware (which is alot)!!!

also it is a full IMU yes similiar to helicommand, but once again modular, and alot cheaper!!! and you can develop more on your own too past what we have done.

so this is a tinkerer's solution.

but also a full featured solution offering turn-key operation (for you non programmers).

:YeaBaby::YeaBaby:

[corvette321]

we are releasing the ligher version OSD 1st... so well see what type of mixing we offer when we start coding for the full IMU version.

but the full IMU version will be released prior to code completion as a hardware platform for developers!!!

and that will sport the new PIC 32 processor at 120 Drystone MIPS. (~80 mhz).

[Hogster]

That's fair enough :) Could one use this system in-line with a Helicommand to offer GPS holding ability? Although come to think of it I've run out of channels on my transmitter so I wouldn't be able to activate/deactive your device! Hmm!

Keep up the great work! I'm following your progress closely :)


David :)

[corvette321]

well you could use the wireless Xbee 2.4 ghz (or 900 mhz upto 20 miles) telemetry module ... it is 2 way communication.


you could set it up to control upto 8 servo lines on the board... , or digital in / outs.

whatever you like...

to control a camera, to switch on an RX mux or similiar....

etc..

:thumbup::thumbup:

[corvette321]

:thumbup::thumbup:code code code writing!!!

got alot of features done already..


everything is in floating point!!! no interger rounding... so have the most accurate possible values....


-altitude in feet and meters upto 99999 feet or meters....
- AGL and MSL selectable (AGL is latched upon meeting threeshold values of minimum HDOP , VDOP , PDOP, sats #
- lat and long is ddmmm.mmm and decimal degre with high accuracy
-speed can be mph and kph upto 999 :) with decimal point for slower speeds ie: 0.xx 00.x

distand to home and distance to target based on lat, and long are done using great circle forumals with high accuracy floating point....

DTH and DTT displayed in same units as others selected...

also whats cool is when you go below 1.0 unit (1 KM or 1 mile) the DTT and DTH automatically switch to feet and meters.... pretty cool feature :)

its super stable right now....

still have some left to do:

--addon modules

1) current sensor and mAH meter
2) SCP1000 baro altimeter module code
3) magneto or compass code instead of COG parsing
4) compass bar.... almost done actually.
5) other addon modules.... rpm?

so its getting there!!

really happy with the eb 85 a's accuracy!!

:thumbup::thumbup::thumbup:

[Neil_J]

Very cool. I was going to design an IMU board very similar to this about a year ago but I've been too busy. Looks great so far. Thanks for keeping the project open-source, I for one will be contributing my work to the project.

My only negative comment would be the processor choice. I was a bit bummed to see a dsPIC/PIC32 tying everything together -- I would much rather see an ARM9 (e.g. LPC3180 (http://www.standardics.nxp.com/products/lpc3000/lpc3180/) or i.MX31 (http://www.freescale.com/webapp/sps/site/prod_summary.jsp?code=i.MX31&nodeId=0162468rH311432973ZrDR)). This is just my opinion so don't take it the wrong way. I first learned to program on PICs years ago, which put a bad taste in my mouth. It wasn't until I moved to the ARM instruction set that embedded programming actually became fun. A favorite quote of mine from ladyada (http://www.ladyada.net/library/picvsavr.html):
Overall, I've written some asm for both processors and I hate to break it to PIC users but writing assembly for PIC is akin to stabbing myself in the face. (Except its not even that efficient, cause you have to more the knife into the working register first (movlw KNIFE), and then you can stab yourself (movwf FACE).)
At least the newer PIC32's are MIPS M4K core and have JTAG. Much better than their previous (proprietary) cores and programming interfaces. So I'm sure the PIC32 will do great. Wonderful, I get to learn another CPU architecture, another instruction set, and another build toolchain :roll:

[corvette321]

Hi Neil,

the dspic is only for the light OSD version :)


the pic 32 is a drop in for it.... (ive already looked over the pins and have a pile of the pic 32;s right here waiting to be installed!!!)

ive been rigorously testing out the OSD code..... fixing bugs...

i like stable code!!

did some driving over to bradenton today over the bridge, and was observing a small 7 inch LCD i have hooked upto it./.....

fixing a small bug in the bearing calculations...

the distance calculations is nice.... making it an option to switch between haversine and regular pythagorean... i want to include vicenty 0.5 mm accuracy distance calc code too but ill have to see if the processor can handle it as it is pretty intensive!! but super accuracte...

i have been getting accuracy from 10-25 feet depending on how many sats i have,.

today i had 11 sats!!!!

gonna bump the OSD upto 5 hz next and see how it handles.... but the code is coming along nicely....

youd be surprised how mucht the dspics can do!! remember it has DMA!!!

all the A2D can be handled in DMA.....so can the SPI.....

thats how you can fit a complete autonav on a dspic!! good code writing.

thats why i refuse to release it until i have all the bugs worked out... and will test it rigorously and completely.

also im packing it with tons of features...

but yah the light version is almost ready.....code is coming great....

i am happy with the COG parsing and compass scroll. it is working pretty well.....(just gotta fix 1 bearing bug.... no biggie...)

tested it out to 50 miles today and its working great....

next hooking up the altimeter SCP1000 sensor and the current sensor and code that.. A2D will be running in DMA mode so less processor utilization. SCP1000 is SPI so thats easy to do...easy code to write...

the 2 axis compass is primarily for those wanting a compass on the ground (slow moving robots etc....) whereass the COG will not work!!!! well it works well when you are moving but for really slow speeds it doesnt give accurate turns etc.,.. thats where the 2 axis compass module support will help out. it is not intended for the air.

maybe further on down the road will do the 3 axis tilt compensated compass readings from the 3 axis magneto data.... that shouldnt be too hard to do, but of course that requires the accelo data....

decided not to include the 3 axis accelo with the light version, as i think most people will use the COG from the GPS....

btw: the pic 32 is a nice chip.. i wish it had a FPU or MMU.. but thats ok... 120 DMIPS is still nice!!!

they have USB versions coming out.

the next full IMU and OSD will be a bob 4 unit with dual pic 32's.

-one pic 32 for sensor / OSD IO

-one pic 32 for kalman / servo driving....

both communicate via SPI at 10 mhz or soo.

that should be around december.

well see how this one sells 1st!!!

:lol::lol::lol::lol::lol:

[Neil_J]

btw: the pic 32 is a nice chip.. i wish it had a FPU or MMU.. but thats ok... 120 DMIPS is still nice!!!
i.MX31 does 740 MIPS, that's over 6 times faster :shock: It has DMA, MMU, vector floating point processing, MPEG encode/decode, CCD camera interface, etc.. I'm not too worried though, I will be using an ARM to do the important stuff, the dsPIC/IMU board will be its slave :DH
thats why i refuse to release it until i have all the bugs worked out... and will test it rigorously and completely.
It would be a good idea to let a few people look at the code before the first release, i.e. peer review. That way anything that gets missed will have a chance to be fixed. I will volunteer to look everything over if you want. I'm somewhat of a GPS guru so I'm sure I could help out.

[corvette321]

hahahah yah freescale imx is nice...

we are playing around with cirrus logic 9315 right now....

(trying to find some applications for it that will be profitable...)

still even the cirrus logic doesnt touch the imx.....

had fun compiling win CE code!!!!

wow that takes a long time even on a dual processor system hahaha....

but yah you could have a nice little embedded linux autonav hahaha!!

but i think thats overkill for a heli or a small rc plane.

maybe for a predator aircraft.

:banana:banana:banana:banana:banana:Bang:Bang:Bang :Bang

[corvette321]

sure you could use it purely for a sensor / OSD board...

it has SPI to pins... so yah slave it all you want!!!!

ill be proud to say its in your autonav system.

:noteworthy:noteworthy:noteworthy

[corvette321]

hahahah found the error in my bearing algorithm...

forgot the %360 at the end.

how silly of me!!!......

:shock::shock::shock::shock:

[r40734]

hahahah yah freescale imx is nice...

Buy the Freescale stuff (keep food on my table). :YeaBaby:

This is an incredible setup you are designing. :shock: From all the features you have mentioned already, and being open source and modular, really got my brain working on the possibilities (scene recognition, fly-to locations, etc. hint hint)

Keep up the good work :thumbup:. I'm anxious for this to hit the market and see just what everybody comes up when they really start playing with the possibilities.

[corvette321]

glad i got some other people excited about it...

also including a GPS logger feature on it.

the windows GUI portion of it will format the data for google earth .KML files.!!!!!

also for google maps.

working on the windows GUI setup program right now.

then just finish a few addon modules and the manual and this baby will be released!!!!!

:thumbup::thumbup:

[corvette321]

well have a flying demo video up after the first week of may!

it wont be a heli though, as we havent gotten our heli platform together yet.

itll be on a foamie as we will try and do simple control algorithms on it...just to start with!!

and well be going to 5 hz gps rate that week also.

hmmm so lotsa things going on here.

:YeaBaby::YeaBaby::YeaBaby:

[Neil_J]

also including a GPS logger feature on it.

For the next version you should include an SD card interface, and write the raw NMEA data to text files. I have the Sparkfun GeoChron (http://www.sparkfun.com/commerce/product_info.php?products_id=8301) logger that works like this. Really convenient to use.

Also no need to stop at GPS logging, acceleration data, altitude, etc could be written to the card if the user so desired.

[corvette321]

great idea,

perhaps with the pic32 with usb OTG......

as soon as the light version is released, will slap a pic 32 on and open her up a little hahaha :)

theres a pic of the setup utility that shows some of the options...

still adding more!!!

the GPS logger utility will be a separate program.

u can see a pic of the setup utility to see what options you have here

http://www.thesiliconhorizon.com/motion.htm

[corvette321]

im thinking the 2 axis compass can be used on helo setups where the speed is less than a cutoff amount...

just driving around today doing more tests, the COG updates from GPS going really slow about 1-2 mph but not below that obviously...

thats where the optional 2 axis compass comes into play.

yes a 3 axis with compensation algorithms is best.....

a 3 axis compass off the shelf is expensive (with tilt control)

therefor a 3 axis magneto can use that an accelo to computer tilt compensated headings all the time...

but for cost sake, (and processing power) and since the GPS handles it over a certain speed, i think the 2 axis compass with a switchover will work well as a solution.

the 3 axis option will be for the full IMU version.

[corvette321]

distance 2 home and distance to target now blink when under 50 feet or 15 meters.

its the little details that count!!

:clappp:clappp

[furyphoto]

I am trying diligently to follow, but the programming/tech details are a bit beyond me. I'm not so much interested in logging, as I am in control. How (in techno idiot terms) will the unit be able to integrate, and help me to control my AP heli?

I think most AP'ers are interested in hands off altitude and position hold so that they can turn their focus (no pun intended) to the the camera operations for a few minutes. Eventually I am interested in a full autopilot system, with gps waypoint flightplan programming to complete, for example, regular inspections of cell phone towers, or construction progress shots taken from the same aerial position every week.

It's sounds like eventually this is the direction you are headed with external add on sensors, Can you please outline in layman's terms the ultimate abilities of you project relating to Helicopter Aerial Photography.

I might be your first customer! but for some of us, pic32 is what you do to your nose after pic31.

Thanks,

-A

[corvette321]

hahahahahah :cheers:cheers:cheers

well pic 32 is a new Microcontroller, and it has alot more processing power to run the kalman filter that powers your autonavigation device.

coincidentally i am working on the altitude SCP1000 addon module today....

well see how accurate it is! the data sheet says it can read pressure Pa accurate to 0.2 meters tranlsated to altitude, but with temp error??...

hmmm

i think it might be accurate to a foot or so depending on the math magic.

well c.

yes for autopilot will do heading holds, circles, alt hold, hover. that is the goal. will do on several diff platforms...

the programming is interesting!!

it is actually good to know what equations the device is using since it will produce errors.

if you go buy a device that uses inaccurate equations and you trust that device for navigation then watch out!!....

regular triangular for distance calculations can have upto 10 percent error!

haversine (spherical) can get you within like 3 meters or soo (0.3 % error)

then theres the ultimate vincenty equations that require lotsa processor power and are good to 0.5 mm !!!!!!!!!!!!!!!!!!!!!

so if i pick a spot on my map to recon and fly FPV to it to see that hidden Incan temple (hahahaha) then i hope my equations are accurate considering my distance is 4 miles out.. (10 percent of that is almost half a mile in inaccuracy!!!!!).

so depending on what you use it for will depend on how much you gotta know about it.

but i like to provide the fullest details as possible as not to sugar coat or limit any possibilities.

:cheers:cheers:cheers

[Tonystott]

Gee Corvette, I think you have to increase the gain on the "plain English" analog control a bit more! LOL I too am trying very hard to gain a good understanding of your development, which certainly sounds fascinating, but it is hard work :)

[Neil_J]

You are planning on implementing a Kalman filter, right?