Error while Integrating Acceleration to get Position after Fusion

[engtec]

Hello,

I need to get position and track the motion of the IMU from the acceleration and other data it gives. I am using a Nucleo STM32F401RE board and a X-Nucleo- IKS01A2 MEMS sensor. I am using the DataLogFusion example in Unicleo-GUI. To get position I am integrating the Linear X, Y, Z acceleration output twice.

For even small movements lasting just for I am not able to get even a reasonably ok plot of the motion path of the sensor( for example when moved in a circle on a flat surface , after integrating I get something else). I have observed that there is a general parabolic increase in position which overshadows the actual motion in most cases.

A few insights on how to solve this problem and accurately get position from the ST IMU ( LSM6DSL) after Fusion would be of great help to me.

[Miroslav BATEK]

This is very difficult scientific discipline, it is almost impossible, because the sensors are not ideal (absolutely precise, without noise, fast, ....). You can do some improvement like filter value, ignore small values (noise), null velocity, ... but I think you always will be disappointed from the result.

[engtec]

Ohh, Thank you for your insight.

I was under the impression that after post-processing the data well with a few algorithms it would give a reasonably accurate estimate as this seemed to be a quite common application. I will implement null velocity, ignore small values and check the improvements.