which Gyroscope IMU recommended (own signal processor, no compass)

Other Hardware Sensors
1

hello,
which Gyroscope+Accelerometer IMU can be recommended, feat. an own on-board signal-processor (like the CMPS11, but no compass for sensor fusion because of many ferromagnetical robot parts).

Nevertheless, it should be as simple to poll the filtered and fusioned values like for the CMPS11,
no filter code and no interrupts to be run on the Arduino itself,
so that the code can be expected to work both for an Uno and for a M0 or Due as well
(e.g., just poll 2 registers for filtered+fusioned yaw values) ,
and actually only 2D yaw/heading is needed, no pitch and no roll:

  // CMPS11 example for comparison
   Wire.requestFrom(0x60, 4);         
   while(Wire.available() < 4);            
   uint8_t HdgHibyte = Wire.read();         
   uint8_t HdgLobyte = Wire.read();     
   double heading = ( (double)(HdgHibyte<<8) + (double)HdgLobyte )/10.0;      
   //int8_t pitch = Wire.read();             
   //int8_t roll = Wire.read();

Who has got experience with such a IMU sensor by himself?

2

A magnetometer is required to correct the drift in the yaw angle.

The BNO055 self contained IMU works very well, but you do need to follow the calibration instructions after mounting it in your robot. That will usually correct for local distortions of the Earth's magnetic field.

If not, mount it in a different location on the robot.

3

as stated, I already have the CMPS11 feat. a 9D IMU including a compass, but the compass inside is not usable because of many ferromagnetical robot parts. I would need >1m distance to make the IMU work. So I need an IMU which provides on-board sensor fusion and filtering without a compass / magnetometer.

4

As stated, your idea won't work.

the compass inside is not usable because of many ferromagnetical robot parts.

This is because you have not properly calibrated the magnetometer while mounted in the robot. See this overview.

5

Compass calibration would not work, as there are also many widely moving metal parts (e.g., a big iron robot arm on the top plus varying metal things picked up by the claw and loaded to the load area). I know compass calibration and compensation for airoplanes and ships, but all that would not work any more as soon as the robot arm starts moving and loading ferromagnetic things additionally. So a digital gyroscope is needed, fusioned with an accelerometer and whatever additionally and out-filtered gyro drift - but no magnetometers.

(btw, why does one have to wait 5 minutes to correct typos in one's post?)

6

If it is absolutely impossible to use a magnetometer, some other external reference is required to eliminate gyro drift in the yaw angle.

The "5 minute" rule will go away after a certain number of posts. It helps prevent people from spamming the forum.

7

If it is absolutely impossible to use a magnetometer, some other external reference is required to eliminate gyro drift in the yaw angle.

actually that is just what my question is about.
Which Gyro will work with what kind of sensors to be fusioned to out-calculate gyro drift, except magnetic field? Perhaps Koriolis force?
After all, I actually doubt that magnetometers are indispensible for IMUs , e.g. check this:

Anyway, perhaps someone already has good own experiences with such a thing.

8

Of course you could use a standard gyroscope, rather than a rate gyro as found on common, inexpensive IMUs.

Standard gyros intended for accurate navigation tend to be large and quite expensive ($10,000-$50,000).

9

are you serious or are you kidding? my question is about an inertial digital Gyro, and did you watch the youtube video? I don't hink you are right claiming that a magnetometer is indispensible for that purpose. Perhaps someone else could answer who as experience about that task or topic.

10

I'm serious. Check this page out.

Rate gyros measure the rate of rotation, not the rotation angle.

Because all rate gyros drift and the drift depends on the temperature, it is mathematically impossible to measure an absolute orientation with a rate gyro, without using at least two external references.

Usually, those references are North and Down, but the references could also be the location of two fixed points in the environment.

Good luck with your project!

11

I know that digital Gyros measure rotation speed, not angle, but angle is just the integral of speed over time, so that would not matter. Also temperature is no issue, just measure temperature and fusion that with the Gyro model. And also in your reference I see just explanations about how common industry sensors work by which sensor fusionings and about common ways of calculations and filters, but I don't see a mathematical prove (e.g., a contradiction prove) that it's mathematically impossible to get a stable yaw reading without having a compass additionally.
And again, please also check the youtube video, appearantly they just use a gyro and an accelerometer for stable 3D position, heading, and path calculation.
edit,
and the "natural" gyro drift caused by earth rotation is a constant, i.e. 360°/24h = 0.0042°/sec, so as a constant this part also can be easily eliminated.

12

as posted bevore, you need somthing to eliminate the drift.
There is one other possibility for exact heading, but only under free sky:
Use two RTK-GPS Sensors. (ie from drotek, aprox 230€ per pice)