BNO055 Calibration & stops responding in forward motion Moderators: adafruit_support_bill, adafruit

[gammaburst]

In NDOF mode the BNO055 sets the gyro range to +/- 2000 degrees/second, but as I rotate the sensor in my hand faster than about 200 degrees/second on any axis (somewhat dependent on compass heading), the cal status frequently drops from 3 3 3 3 to 2 3 3 3 or 1 3 3 3. I don't yet know if the output data is actually degraded or if those cal indicators are merely false alarms. Interestingly, the cal status remains 3 3 3 3 during vigorous shaking (mostly linear acceleration) even though the output data is severely clipped at +/- 4g. My hunch is the BNO's firmware isn't very smart. Also, the auto-calibration may briefly disable itself during moderately fast rotations due to the magnetometer's relatively low bandwidth (10 hertz) and sample rate (20 hertz) - just another hunch. Someday I'll try comparing the BNO's output to a high-quality MEMS sensor (with three more digits in its price tag).

[peyopeyo]

Same here.
I will change to change de gyroscope range, maybe you're looking in the right direction.

Edit : you cant change the range in fusion mode as it is auto controlled.

[stefdas]

hello pauleilo,

I finally receive the MPU9250 and get some time to test it.
running on arduino mega, slerp value at 0.005 and testing it in my car.
Is is far better than bno055 but acceleration still influence a lot the titl/roll measures.
i'm a but disappointed....

regards

stef

[pauleilio]

Hi stef,

I recently noticed the acceleration influence. Would it be possible to remove a percentage of the raw acceleration from the fused output with a bit of code. I haven't tried it myself due to the fact that acceleration does affect an aircrafts attitude indicator (so I am led to believe)

[skypuppy]

Aircraft attitude is my primary mission for this beast, so the topic is really important to me. My thoughts on the errors are:
1. using an int where we need a long or a float,
2. maybe the internal code is doing #1,
3. the autoranging is not working and we're exceeding an int value when we get above 6 mph or so,
4. maybe we not initializing it correctly,
5. or it's broke. :)

[stefdas]

hello Pauleilio,

that's the purpose of the "slerp" parameter.
While testing in my car (up to 60 km/h) using a value lower than 0.005 gives really bad results; a lot of drift, so I assume that acc part is too tiny to compensate gyro drift.

Except acc influence in curves, the good news is that measures are stable over time and over speed (up to 100km/h in my car); I run the test during one hour.
In my application, this may be not have a big impact because the vehicle speed is low (tractor field's work at max 20 km/h), and very "sweet curves" are executed while working in the swath. However I should place the sensor as low as possible in order to minimize sensor displacement due to frame/cab bouncing.
I will made some tests in July when crops will be done.

regards

[pauleilio]

Hi Stef,

Yes, I noticed that 0.005 was the absolute minimum it would go before drift crept in. I tried changing it in runtime but that didn't work. My device is away at the moment so will try with it when it gets back.

Hope your test's prove successful in July.

Still not able to get hold of a BNO070 yet. Hillcrest offer a very expensive dev board.

[sylvainmahe]

Hi guys :)

I bought the BNO055 recently (for computed pitch/roll/yaw outputs), and I test this into my quadcopter:
https://www.youtube.com/watch?v=3daBac7hJnM

I use BNO055 with my own library: http://sylvainmahe.xyz/
& main code: http://sylvainmahe.xyz/cpp/

I running in NDOF mode.

When I make a continuous translation (exemple: go forward), with my quadcopter, 0 degree of pitch axis is no longer horizontally !

Look my video at 3:15 when I move forward/backward:
https://www.youtube.com/watch?v=3daBac7hJnM&t=190

It takes a few seconds of static quadcopter to find a real 0 degrees (horizon) !

I think it is because the magnetometer loses its calibration... (I use autocalibration at powerup).

...Furthermore, I am forced to use the mpu6050 gyro because BNO055 gyro at 100hz (or even less) is not fast enough for good locked quadcopter on 3 axis.... (mpu6050 have 1000hz gyro outputs !)

BNO055 :( :( ....

[sylvainmahe]

For 0 motion (""stops responding in forward motion""), it's strange but:

-> When I move BNO055 in my hand (without metal screws), BNO055 randomly back to 0 values outputs...
-> When I move BNO055 in metal frame (fixed with 4 metal screws), BNO055 don't back to 0 values outputs, it work all the time...

BNO055 = LOL DEVICE

[sylvainmahe]

AND it's not RESET pin, because this pin have pull-up resistor, and if I put +5v into this pin, BNO055 randomly back to 0 too !!!

[pauleilio]

It is actually a shame the BNO055 doesn't work, it would have been a great device.

Lets hope their BNO070 is a lot better.

[skypuppy]

Dang, and I have FOUR of them!

[sylvainmahe]

Bosch engineers come here !!!

[skypuppy]

Boschmen, come out, come out, wherever you are.

[sylvainmahe]

Ok guys, I tested yet BNO055.

-> I created C++ code to calibrate and watch CALIB STATUS of BNO055.
-> I calibrate BNO055 in NDOF_FMC_OFF (slow calibration = more precision)
-> I created C++ code to send CALIB STATUS value (by RF) air sol

-> Ok, my BNO055 on my quadcopter have byte 11111111 CALIB STATUS throughout the flight !

AND.......

The same !!!

-> When I make a continuous translation (exemple: go forward), with my quadcopter, 0 degree of pitch axis is no longer horizontally ! Like my video at 3:15 when I move forward/backward:
https://www.youtube.com/watch?v=3daBac7hJnM&t=190

It takes a few seconds of static quadcopter to find a real 0 degrees (horizon) !!!!


Bosch engineers HELP ME, I DON'T UNDERSTAND YOUR DEVICE !!