Iterative learning control - HeSunPU/grid_slam GitHub Wiki

Through observation, we can find the mobile robot always deviated to the left direction while moving, which means there exists some errors in repetitive mode in our odometry measurement. As Kalman filter is effective in estimating zero mean noise of Gaussian distribution, we need to find new methods to solve the systematic problem. Iterative learning control is a learning process, which is always used to fix the repetitive systematic errors.

reference: Bristow, Douglas, Marina Tharayil, and Andrew G. Alleyne. "A survey of iterative learning control." Control Systems, IEEE 26.3 (2006): 96-114.