We have prepared all the necessary materials for acquisitions:
- encoder of the two wheels
- gyro velocity output
- a reference system ground truth to compare your algorithm and eventually optimise parameters
Send an email to alessandro.luchetti@unitn.it to schedule a day to come to the lab to acquire the data.
You could follow two different approaches:
- Sensor Fusion in the statistical domain, i.e. Bayes or Kalman filtering
- Sensor Fusion in the frequency/time domain:
- fusing only attitude increments with 1-dimensional Complementary Filtering
- fusing also position. In this case use quaternions
