Path reconstruction based on gyroscope bias estimation using GPS

This paper presents a method for path reconstruction from logged MEMS gyroscope and odometer measurements, through the estimation of the gyroscope bias from GPS data. The bias estimation is done by minimizing the distance between the reconstructed path and the external reference positions (GPS) by tuning the bias with common minimization methods. Both gradient based and quasi-Newton minimization methods are investigated. An efficient quaternion based mathematical model of the problem is presented. The path produced by the proposed method has higher spatial and temporal resolution than the GPS path, it is more accurate than a path from a purely inertial measurement unit and it is at least as accurate as the GPS path, according to real-life experiment.