A kalman-like filter based on joint discrete and linear-Gaussian models
X Merlo · 1st IASTED International Symposium on Signal Processing and its Applications · 1987
In digital signal processing techniques for detection systems or for robotics, Kalman filters are often inadequate, although very efficient when they can be applied. This is partly because the perception has also to deal with discrete states of the universe and discrete observations (recognition results..) . In the past years, there were advances in the estimation and control theory for linear models with jumps between discrete parameters (Caines and Chen, Willsky and Jones). The availability of fast and powerful DSP microprocessors allows us to use such complex stochastic system models. We propose here a similar parametrization for linear-gaussian markovian models in a discrete time.