Skip to Main Content
The author comments on the work of X. Rong Li (ibid. vol. 33, pp. 1212-24, 1997) in which it is mentioned that since the kinematic models are completely observable and controllable, the Kalman filter is stable and will converge to its steady state with a constant filter gain. Alternatively, a simpler implementation is to use directly a Kalman filter with the steady-state gain from the very beginning. It is contended that these statements are debatable. Rong Li replies to the comments.