Towards Robust State Estimation by Boosting the Maximum Correntropy\n Criterion Kalman Filter with Adaptive Behaviors

This work proposes a resilient and adaptive state estimation framework for\nrobots operating in perceptually-degraded environments. The approach, called\nAdaptive Maximum Correntropy Criterion Kalman Filtering (AMCCKF), is inherently\nrobust to corrupted measurements, such as those containing jumps or general\nnon-Gaussian noise, and is able to modify filter parameters online to improve\nperformance. Two separate methods are developed -- the Variational Bayesian\nAMCCKF (VB-AMCCKF) and Residual AMCCKF (R-AMCCKF) -- that modify the process\nand measurement noise models in addition to the bandwidth of the kernel\nfunction used in MCCKF based on the quality of measurements received. The two\napproaches differ in computational complexity and overall performance which is\nexperimentally analyzed. The method is demonstrated in real experiments on both\naerial and ground robots and is part of the solution used by the COSTAR team\nparticipating at the DARPA Subterranean Challenge.\n