IMU intrest

Has anybody looked at this,

as its open source, maybe GHI could take a look.


Looks very good. Thanks.

Looks like a good start. It needs on-board filtering (ARHS) to get me excited about using it with NETMF. Also, while you’re taking notes Gus, an idea I have would be to also add a port where an ultrasonic rangefinder could be connected to supplement the low altitude data so that in a copter you only have to be concerned with one altitude reading instead of two.


There is an AHRS in the software package which is easy to port to NETMF, but which type to use can be a touchy subject with RC pilots. They like to stick (no pun intended) with what they see works for them, and who can blame them, when a failure can hit the ground hard.


Sure you could build the ARHS in NETMF, but the advantage I see to having it onboard the IMU is that you get the realtime benefits of a separate mcu that can be dedicated to reading many measurements and doing the filtering w/o stealing processing power away from the NETMF mcu. Thus, leaving more cycles and memory for complex navigation rather than figuring out the basics of balance and direction. I don’t think most building R/C’s using NETMF/Gadgeteer boards will be your traditional R/C pilots :wink:

I had a feeling thats what you ment. The latest iteration of the IMU at the above site uses the mpu-6050 which does have that fusion algorithm., but there are two big problems,

1 The magnetometer that Invensense paired with the device is expensive.

2 You need to have an NDA to find out how to use or turn on the said algorithm. (very bad)

Having said that the cheap HMC5883L can be attached to the mpu-6050, so you can save a few ticks. Maybe a future release will allow any magnetometer to be used with their fusion algo. which would make an execellent solution.

Just a quick edit checkout the new but unreleased mpu-9150 your dream come true @ ianlee74.


My dream would only come true if Gus would take that chip and turn it into a Gadgeteer module with GHI support :slight_smile:

BTW, this is what I’m using now. Good product. Bad support.

@ Ianlee74
Its amazing what the Atmel chip can accomplish. I am also a little puzzled why some one with your experience would have the need for support, anyway one question, what bandwidth does the main loop have on your Mongoose?

@ Gus
The MPU-9150 has a release date of 4/16/2012 but dont hold your breathe the MPU-6050 was a year late! There maybe free samples available, and just think GHI could be first to utilize the device. :wink:


I’m a software guy first. Believe me the reason you see me here is because I have more questions than answers :wink:

I’m not sure… I haven’t dove into trying to mess with the firmware yet. I’m still using the stock firmware with 115200 baud serial. Converting it over to SPI was going to be my next steps. mammaplank has been working on this already and might know more.

$25 in volume! I thought their press release said low cost :slight_smile:


If you sell a module for $60+ it would still be the cheapest IMU out there! (I think)

25chip+board+distributor = we lose money at 60 :slight_smile:


It was worth a try!
On a much more serious note I would suggest getting a few samples to try, so you are sure the package will fly a multi-copter. Im sure the price break is high, so a big commitment. I am aware that there are a number of non flight applications, just a thought… Thats if your going down that route. ::slight_smile:

How about if you just add one to a copter version of the Cerb40. Then at $60 for the whole thing you’ll be just fine :wink:

Correct me if I’m wrong but, there isn’t a shipping Invensense hobby IMU board that enables the on chip engine. They only have example code for DCM, MARG or enhanced Kalman filter running on an ARM, Arduino or PIC. Seems Invensense has no interest in supporting the hobby market. Can GHI change this? World be a good fit for these products. The new Sparkfun 6050 board has a library posted in the comments that does load and run the engine on Arduino.

IMU is already in the works. The goal is, very professional output but at a reasonable cost.

Anyone can put accelerometer and gyro on a board, the callange is in having a decent output.

I have try during a week to obtain a reliable freeimu solution with the 10736 from sparkfun. Honestly even if the work is really great ( taking into account that software is flexible and covers a large set of usage) I have not been convinced by the core algorithm. I really like the startup calibration phase even if I think for the 10736 magnetometer seems not use negative calibration bias… (still have some doubt). The major trouble I found is that for the 10736 from sparkfun, the cube seems to have a delay to converge into the correct position. Modification of the Ki Kp constants can improve it but its as usual a tradeoff between accuracy and latency…

I did not have chance to by the freeimu HW and I hope this is better (it seems yes when we look on video).

Actually the best one that I found but that is really really really (… and may be too much) expensive is the X-IMU…

For now the most reliable solution is the razor 9dof solution. Do you have other product to advice me?

@ leforban,

The “best” IMU’s are the ones that you as a pilot are happy with, but more commonly what a group of pilots are happy with. Most hobby IMU’s are based on the itg3200, adxl345, and the hmc5883, so its all down to the software solution. I think I am safe in saying that any good 3D pilot would easly out fly any of the current hobby IMU’s, then IMU’s are mostly used to stablise flying platforms, its only recently ( last 5 years maybe longer) IMU’s have been used to find way points, with the introduction of a GPS. So the “best” in real terms is what you can afford. The only concern I have with these new “all in one” solutions is the data rate.

Hi I am a little new to all this.

I have been building an IMU 9DOF AHRS using a FEZ-MINI and a Pololu MinIMU-9 (at $50 is about the cheapest I have seen). The Arduino environment really really sucks, and dont want any part of solution dependant on Arduino so I ported the Arduino code to C# successfully (mostly, I will come back to that). Total cost $70 but thats mostly the Pololu board.

I have the FEZ Mini outputting Roll, Pitch, Yaw serially and I have the Python graphical display app showing the results. So far so good. My biggest problem is I really dont fully understand the maths yet and I have a few calibration issues that means the board orientation does not quite match the data, through apart from that its doing the right things.

Anyway I am close to being able to post what I have in the hope that better minds than mine can help improve on this. I will swap out the Python app with a nice .NET one too. Has anyone else attempted a pure NETMF 9DOF AHRS solution and got good results? Dont see there should be any huge hurdles to this.


Very cool. Can you share the code on code share when done?