A typical low-cost MEMS-IMU with 6 DOF (3-axes accelerometer + 3-axes gyroscope) might not be able to solely keep track of orientation and might diverge due to large out-of range acceleration peaks (caused by mechanical shocks).
I thought about something like an adaptive Filter (e.g. error state KF) which recalibrates the IMU on the run or some smoothing algorithms which postprocess the sensor data to compensate irregular data caused by high dynamics.
Are there any related papers which somebody can recommend?
Cheers
Andreas
Caruso, M.; Sabatini, A.M.; Laidig, D.; Seel, T.; Knaflitz, M.; Della Croce, U.; Cereatti, A. Analysis of the Accuracy of Ten Algorithms for Orientation Estimation Using Inertial and Magnetic Sensing under Optimal Conditions: One Size Does Not Fit All. Sensors 2021, 21, 2543. https:// doi.org/10.3390/s21072543
Monitor the amplitude of the acceleration vector and when the acceleration is 10% larger or smaller than g0 value, just discard the data and interpolate using gyroscopes. Gyroscopes are quite resistant to shocks. After you get normal values of acceleration you can correct any small errors due to pure integration. You can use KF or any other method.
Janez Podobnik Can you recommend any paper which manage this kind of challenge when the IMU goes into saturation? I´ve just found a lot of papers that examine high dynamics (via KF, compelemntary filter, etc.) but never answer the question about saturation conditions and how cross sensitivity of the gyros might influence your mentioned interpolation during high shocks. Kind Regards
Andreas Siebert Maybe this paper https://ieeexplore.ieee.org/document/9164809 (Adaptive Complementary Filtering Algorithm for IMU Based on MEMS).
What i was proposing is purely practical approach, there isn't much theory behind. It is part of all the purely practical engineering if-then conditions that improve the quality of estimation, but are usually not included in papers.
Janez Podobnik I see. Thank you, indeed this work seems more to be the kinda promising approach which I´ve been looking for.
In general your mentioned practical proposal just to discard the data duing high shocks is reasonable but I was wondering about the cross sensitivity of the single axes. For example the influence of the gyroscope measurement during a mechanical shock in one direction though a gyroscope is more shock resistant against linear acceleration.
Andreas Siebert When it comes to resistance of gyroscope to high linear accelerations, they are pretty resistant. Usually documentation gives you sensitivity to linear acceleration and now days all gyroscopes are pretty much insensitive to linear accelerations. At least compared to all other disturbances that you will have to deal with in practical applications.
- Badges
- Science topic
- Similar topics
- Control Systems Engineering
- Filtering