Wrong magnitude of linear acceleration from the IMU

Issue #63 on hold
Dimitrios Kanoulas created an issue

Hi all --

it turns out that I am getting wrong values from the IMU sensor at least regarding the magnitude of linear acceleration. To test it please place the IMU statically somewhere and with even the mag and the gyro off, measure the output values of: /multisense/imu/imu_data

I am getting something like: linear_acceleration: x: -6.74697542913 y: -0.0784532037263 z: -8.55139917877

Which is only from gravity and its magnitude should be 9.8, but it is around 10.8. Any idea why?

Btw I tried to produce the IMU orientation using the madgwick filter, but having lots of problems. I assume one reason is also the wrong values coming from the IMU. Any hope to have an integrated orientation in the FPGA?

Thanks a lot!

Comments (2)

  1. Chris Osterwood


    We have seems similar zero-motion accelerations from the LSM303DLHC accelerometer -- it does not have very good default calibration from STMicroelectronics.

    Scaling the acceleration down to expected values prior to the madgwick filter may help it perform better, but we have not used that package so don't know how much that could help.

    When we use the MultiSense IMU in our own projects, we've always run a pose filter off-board on general purpose processors. The Microblaze CPU inside the MultiSense FPGA does not have enough processing headroom to run a good pose filter, and investing the time to port our pose filter onto the FPGA logic has not seemed worth while given the limited performance of the STMicroelectronics components we use.

    I am sorry that I don't have a better answer for you.

    Best regards,


  2. Log in to comment