EKF_CAL: Extended Kalman Filter-based Calibration and Localization

The increasing complexity of autonomous systems operating in challenging environments demands robust and accurate calibration methods.Modern autonomous systems use multiple types of sensors to handle a wide variety of environments.To improve system robustness in these challenging environments, redundant sensors are often used to protect against erroneous sensor measurements or outright sensor failures.However, the underlying navigation system must be designed in such a way as to handle these duplicate measurement streams and the addition of each sensor introduces parameters that must be properly calibrated before these sensors' measurements can be effectively used.Moreover, these underlying calibration parameters can change, and often must be estimated online.Kalman filters are a common and near-optimal method to estimate a system's state given a series of measurements and transitions through time.Improper tuning can lead to filter instability and poor performance.As such, it is important to verify a filters performance and stability across a large set of runs in the desired operating environment with various initial errors in state estimates.