A BNO055 is used with an Arduino Uno and the Adafruit BNO055 library running in 9DOF Mode. Is there a way to increase the accuracy of the fused measurements to get a smaller cummulated error caused by the double integration? Or is there another technique to get better results by using third-party software/components?
Thanks for your inquiry.
BNO055 is able to output dynamic and accurate heading/pitch/roll Euler angles, Linear accelerations, Quaternions and Gravity vectors at 100Hz in NDOF sensor fusion mode. It is not a good idea to use double integration on the linear acceleration to get distance travelled, because any little error on the acceleration will accumulate to large distance error over time.
You may use BHI260AP + BMM150 two-chip solution for dead reckoning. Please see more details at https://www.bosch-sensortec.com/products/smart-sensors/bhi260ap/ about BHI260AP with PDR algorithm built-in. The PDR is based on step counter which will not accumulate to large distance error over time.
There is an article in EDN written by: KAUSTUBH GANDHI AND AMITHASH KANKANALLU JAGADISH, BOSCH SENSORTEC
It shows a very nice result for PDR when using magnetometer recalibration.
I would very much like to acheive these kind of results. Can you give more information about how this was accomplished?