Kickstarter project - Tilty IMU

Found a cool Kickstarter project today called the Tilty IMU. Basically 9 axis motion controller; 3 axis accelerometer + 3 axis gyo, barometric pressure and compass sensor with a microcontroller. 3 UARTs, SPI, I2C, 24 digital IO pins, 10 analog inputs, 10 PWM outputs, voltage monitoring and integrated bluetooth for $60 dollars.


https://www.kickstarter.com/projects/1697625937/the-tiltyimu-a-family-of-user-friendly-robotics-co

I ordered one. Should be fun to play with, especially if you integrate it with an adruino package.

I have no involvement, or relationship with the kickstarter project beyond being a $60 backer.

Why is a kickstarter project needed for that? This is nothing but a sensor board for a teensy.
And 60$ without teensy and bluetooth is not really cheap.

You might want to have a look at the MultiWii project.
If you like to use an arduino you could simply use a FreeIMU board for the sensors. You can get cheap clones on ebay for less than 30$. If you do not want the MS5611 and the MPU6050 you can get a 10DOF IMU for under 13$.

I think it makes absolutely no sense to integrate an teensy board to an arduino.
The teensy can do the same things, and it is faster than the Atmega328.

There are also several all integrated solutions available with integrated Atmega chips, Ardupilot, Mongoose, flyMaple, IOI Mini Multiwii.

I have a FreeIMU - The one's on Ebay you have to desolder and reconfigure to be FreeIMU compatible as the axis are not aligned the same. The cost of one prebuilt is $63 and doesn't have the I/O capabilities of this package with a controller. I've been using it integrated into a Atmega328 to record 9 DOF data for post flight playback. This was a real bear to get working.

I love packages like the Tilty Omni Deluxe (what I ordered)which is really a flight management unit which has dedicated cpu resources sourced for the specifications of the board. I then use an Arduino for my crazy projects which doesn't jeopardize the integrity of the controller - a hard lesson learned after a failed flight.

The closest I had found earlier was a PX4 on diydrones.com which starts at $149 and is still in kit form. If you put up a kickstarter package integrating everything prebuilt at a lower price point, I'll be the first backer at your door :> I'd also like to request GPS integration, a MicroSD slot, 9cm transmitter, and a battery powered base unit with bluetooth to connect to my phone or computer.
[h=1][/h]

I quickly skimmed this Kickstarter page and the dot com domain yesterday, but did not see any real specs and/or performance data for the Tilty... Anyone have some references? Having put together, programmed, and deployed several working ACS projects, I'd like to see how this stacks up against the navigational class IMU's I've used in the past.

MWC said:

I quickly skimmed this Kickstarter page and the dot com domain yesterday, but did not see any real specs and/or performance data for the Tilty... Anyone have some references? Having put together, programmed, and deployed several working ACS projects, I'd like to see how this stacks up against the navigational class IMU's I've used in the past.

Click to expand...

I emailed and asked before I backed it. The IMU is a newer version invensense that's in the FreeIMU. Specs here: https://invensense.com/mems/gyro/mpu-61nx.html . Doesn't look too different then the 6050 version in the FreeIMU, except for higher precision. The three axis compass is this one: https://shop.moderndevice.com/products/3-axis-compass . I have no experience with it at all. The pressure sensor / altimeter is this one: https://www.freescale.com/webapp/sps/site/prod_summary.jsp?code=MPL3115A2 . I have no experience with this either.

Scratching my head over gyro sensitivity spec.... expressed as LSB/°/sec

Well, this is one thing I actually know. Gyro sensitivity is the number of volts change in the sensor reading for every degree per second rotation rate.
IE, 11.4 mV/deg/sec. Numerical output gyro sensitivity is the number of least significant bit (LSB) counted for every degree per second rotation rate.

One of the huge issues I had with the FreeIMU is when you added up all the types of errors (angular random walk, constant bias, bias stability over time and temperature, etc...) I would get error deviation calculations that were higher than the sensitivity of the gyro! This version looks like it handles temperature sensitivity much better, but it's still a consumer device.

I feel like I might be able to clarify some things here, as the owner of the project....
Primarily, the TiltyIMU controllers are not a Teensy integrated with an Arduino. The Teensy is the only microcontroller on the board. And yes, you can get a 10 DOF IMU board for significantly less, but you need a decent knowledge of programming and the math required for a system like that. A large part of my project is the complete lack of required programming for certain projects.
Also, yes my board is similar to a some other IMU controllers out there, but again there's no programming and the Teensy 3.0 is massively more powerful than the AVR controllers used by most.
I'm not saying mine is the best IMU controller out there overall, but mine excels in areas where some of the others fall short, and vice versa. But it's also a brand new product and can only be improved.

And for anyone looking for more detailed specifications, I will look into linking datasheets or at least including more information on my website. Though I can say the gyro at its most sensitive, and the settings I use, is capable of 200 degrees/sec with a resolution of 0.003 degrees/sec (though that level of sensitivity is of course susceptible to noise).

Great to see a reply from the developer.
Actually the thing I am wondering most is, that it states to be open hardware and open source project, but I see nothing of it on the webpage. Why not release the prototype data?

I agree that the idea is nice for people who do not want to bother with the hardware, but simply design a project.
Thats somehow more close to the initial arduino-idea, than the arduino is today.

When considering rocket applications, the problem of all available IMUs is that they normally consider the earth gravitation as the dominant force.

Since you already near 5k$ in a few days, there seems to be a high interest in your project.

If this turns out to reach 1kHz sampling to microsd with a high g-sensor, I will buy one.

Does it really use the MPU-61Nx already? Getting pre production samples from InvenSense would really impress me.

alexzogh said:

Doesn't look too different then the 6050 version in the FreeIMU, except for higher precision.

Click to expand...

I would guess the price will be much higher.
If you compare the MPU3300 to the MPU-3050 its 5 times more.

To clarify some confusion about the sensor, I use the MPU-6050. It's actually reasonably common and is used by a number of other IMU systems (such as the ArduPilot and FreeIMU). And the reason I have not released prototype data is that I want to be sure I have a working version of the final prototype, which I should have within just a few days.
And though I have not tested if it is capable of writing accelerometer data at 1kHz, I know it is capable of reading it that fast (it can do a read of all 3 axis gyro and accel in ~750 us, so just accel should be no problem), and with its much faster than normal SPI speed it should be capable of writing fast enough (some quick research tells me it should be 500+kB/s). I believe the limiting factor here would actually be the accelerometer update frequency on the sensor itself, which is only 1kHz max, while the gyro update frequency goes as high as 8kHz. Also the accelerometer can measure up to 16 G's (so 15 of vertical acceleration) at its least sensitive setting, which I would expect to be enough for most applications.

I can confirm that you should be able to do 1kHz sampling to an SD card. I tested the raw data speeds and reading all three axis accel and gyro on average takes <500 us. And with the SDfat library optimized for Teensy 3.0 you should get write speeds of ~1770 kB/s. Check out this thread for the SDfat data. And I just ran the FreeIMU SpeedTest program to determine the raw data read times, so you could likely improve it somewhat if need be.