Here’s my idea for a ridiculously accurate position-locating system. It would be a board with an LPC2138 or maybe a '2148, which is connected to an HMC5843, a Venus643LP with a Sarantel helical antenna, a LIS302DL, an LY530AL, and an LPR530AL. This would give you high-sensitivity GPS, WAAS, AGPS, 3 axis magnetometer (tilt-compensated), 3 axis gyro, and 3 axis accelerometer. The price of the components alone is about $120. Perhaps Bluetooth or a Nordic could be added to give short range telemetry. The entire thing would be extremely compact and low power (an 1100mah lipo would probably give about 4 hours of battery life), and, best of all, a comparatively low price.
With this flood of new ultra-low cost gyros I am expecting to see a lot of super-tiny single-PCB IMUs coming out on these forums… If you have the money and the time, with those components you/ve selected it sounds like you can make a decent one.
However I would be reluctant to call it ‘ridiculously accurate’ - having worked with many IMU solutions, I can definitely say you get what you pay for. The major challenge will be in the sensor fusion algorithms, and how well it copes with the noise and drift characteristics of your sensors. Do you have any ideas about sensor fusion algorithms yet?
Actually, I like this idea, but with something extra.
Have an onboard microcontroller (AVR maybe) reading all the sensor data and GPS, fuse it together and output all the results over serial. Call it a compete position and orientation system on a board that is capable of Dead-reckoning.
It would give hobbyist (who are not skilled at sensor fusion)the ability to build robots with high orientation and position awareness.
I would buy this for a project i am working on now, even if it cost up to $200.
The main value would be the firmware. The hardware design would be trivial in comparison. That’s why I haven’t bought a sparkfun IMU already, I don’t want to write my own kalman filtering firmware.
check out the vectornav vn-100. its totally awesome as far as embedded AHRS systems. The filtering algorithms are difficult and time consuming to code though. I might consider the $500 price tag if it were open source. With the new 3-axis 4x4x.9mm gyro out it seems like the footprint could be reduced further.
the hardware design of a small format board like this would be simple but the filtering to get the quaternions or a heading matrix out are quite involved and require a lot of tweaking.
an open source product tackling this problem was successful it would be really neat.
To those reading this thread, there is a link in my above link to someone else who has developed a tiny 6-DOF IMU board and is selling it for $125. Looks like Sparkfun parts
Haven't been to RCG for weeks, too many... "smart people" on there :roll:
It just annoys me when people think its 3+3+3=9 like most of the people on DIYDrones, even the most experienced guys keep saying 9DOF… If you get that wrong how are you supposed to be trusted to make a full IMU. Most of these guys are very talented, at copy and paste… (That will come back to bite me for sure!)
Now because of certain “high ranking” (wrong analogy probably) people have said it a few times on DIYDrones, it has filtered down through “the ranks” (to go with the same bad analogy), and now there is a whole “army” (maybe not so bad of an analogy) of people who swear its god 9 degrees of freedom.
Even Dean of Atto fame used to go on about his 9DOF or 12DOF IMU’s he was working on… yeah right, that makes me put a lot of faith in him once he says that, maybe he knows about a few more dimensions I don’t?
Krogoth:
To those reading this thread, there is a link in my above link to someone else who has developed a tiny 6-DOF IMU board and is selling it for $125. Looks like Sparkfun parts
Well, to be fair, not "Sparkfun" parts. I know for a fact he isn't buying them from SFE, and when they are cheaper to buy from Mouser (even for 1pc) why would you :P
gussy:
Now because of certain “high ranking” (wrong analogy probably) people have said it a few times on DIYDrones, it has filtered down through “the ranks” (to go with the same bad analogy), and now there is a whole “army” (maybe not so bad of an analogy) of people who swear its god 9 degrees of freedom.
I’m sure our good friend Mr Kalman would have complaints along similar lines. Although, at least people are being exposed to these kinds of advanced concepts, even if it creates a plethora of questions each week. My favourite being “Can somebody explain the Kalman filter without using any maths?”
gussy:
Even Dean of Atto fame used to go on about his 9DOF or 12DOF IMU’s he was working on… yeah right, that makes me put a lot of faith in him once he says that, maybe he knows about a few more dimensions I don’t?
Are you sure he wasn’t talking about tracking 9 or 12 states? I’ve heard of EKFs that do up to 21 state tracking (pos, vel and accel in each of 6DOF gives you 18 already, plus wind speed etc), but haven’t been game to look any further. I don’t think you really need to estimate all that, you really only need to know a couple of those to accurately fly a UAV.
gussy:
Well, to be fair, not “Sparkfun” parts. I know for a fact he isn’t buying them from SFE, and when they are cheaper to buy from Mouser (even for 1pc) why would you
If you live outside the US, that’s why - here in Aus we get slogged with a minimum $40-50 shipping fee from Mouser or Digikey for anything - even SMD resistors… Free shipping from Farnell and RS though Also my comment was more that pretty much all the parts were available from SparkFun if you want to prototype your own, cheaper, better? one.
Krogoth:
Are you sure he wasn’t talking about tracking 9 or 12 states? I’ve heard of EKFs that do up to 21 state tracking (pos, vel and accel in each of 6DOF gives you 18 already, plus wind speed etc), but haven’t been game to look any further. I don’t think you really need to estimate all that, you really only need to know a couple of those to accurately fly a UAV.
It was "9DOF" and "12DOF", written like that from what I remember, but I could be wrong.
Krogoth:
If you live outside the US, that’s why - here in Aus we get slogged with a minimum $40-50 shipping fee from Mouser or Digikey for anything - even SMD resistors… Free shipping from Farnell and RS though Also my comment was more that pretty much all the parts were available from SparkFun if you want to prototype your own, cheaper, better? one.
Will.
Here in Aus... haha here with me then I guess :)
Mouser is free for over $200aud (yeah I know… $200). I actually find Sparkfun and Digikey shipping prices about the same. I’m finding myself using farnell more and more now, free shipping for a $5 item, overnight freight. Order it at 3-4pm and its there at 7am next morning, an no cost of shipping. Can’t beat that. Didn’t know RS was free too, last time it was $25 or something like Farnell used to be.
Didn’t mean to sound bitter in the last post, had just spent too much at Digikey
(You can find the author’s full thesis online, it’s much more detailed).
Although lacking diagrams (there are a couple of variables that you have to figure out yourself), it goes through the full derivation of 3 cascaded Kalman filters for 7-state estimation (pitch, roll, yaw, latitude, longitude and North and East wind speeds) based on a 3-axis gyro, accelerometer and (2 or 3 axis) magnetometer, as well as GPS and airspeed.
A device like the OP mentioned, plus wind speed detection (pitot) and the cascaded Kalman filter described in the above paper running on a 32-bit ARM, for as cheap as possible… SparkFun are yet to build something with all these hardware features together, but I can’t see it being particularly difficult.
gussy:
Here in Aus… haha here with me then I guess … Can’t beat that. Didn’t know RS was free too, last time it was $25 or something like Farnell used to be. Didn’t mean to sound bitter in the last post, had just spent too much at Digikey
I suspect RS realised they had to match Farnell’s shipping policy, because when you have to pay for both $25 shipping and the ‘RS Tax’ vs just the ‘Farnell Tax’ it’s just not worth it. Where in Aus are you located, gussy?
and look under VIKAS which provides his thesis on “Integration of Inertial Navigation System and Global Positioning System Using Kalman Filtering”
Using 3 axis Giros, 3 axis Accell., 3 axis Magn, Speed and Altitude measurements. The C Source Kalman Filtering programs are easily portable and or adaptable to any microprocessor.
Also my comment was more that pretty much all the parts were available from SparkFun if you want to prototype your own, cheaper, better? one.
… Can’t beat that. Didn’t know RS was free too, last time it was $25 or something like Farnell used to be. Didn’t mean to sound bitter in the last post, had just spent too much at Digikey