Publica
Hier finden Sie wissenschaftliche Publikationen aus den FraunhoferInstituten. A Mnemonic Kalman Filter for NonLinear Systems with Extensive Temporal Dependencies
 IEEE Signal Processing Letters 27 (2020), pp.10051009 ISSN: 10709908 ISSN: 15582361 

 English 
 Journal Article, Electronic Publication 
 Fraunhofer FKIE () 
Abstract
Analytic dynamic models for target estimation are often approximations of the potentially complex behaviour of the object of interest. Its true motion might depend on hundreds of parameters and can involve longterm temporal correlation. However, conventional models keep the degrees of freedom low and they usually assume the Markov property to reduce computational complexity. In particular, the Kalman Filter assumes prior and posterior Gaussian densities and is hence restricted to linear transition functions which are often insufficient to reflect the behaviour of a real object. In this letter, a Mnemonic Kalman Filter is introduced which overcomes the Markov property and the linearity restriction by learning to predict a full transition probability density with Long ShortTerm Memory networks.