**3.1 Bayesian networks**

Bayesian Networks are models which capture uncertainties in terms of probabilities that can be used to perform reasoning under uncertainty. It epitomises probabilistic dependency models that represent random stochastic uncertainty via its nodes (Darwiche, 2009). The Bayesian inference has been widely used especially in the localisation problem (Thrun, 1998), in the mapping and in the learning mechanisms in mobile robotics. But this technique is very slow and complex (NP-Complex) because it uses a probabilistic reasoning that require much more time to choose a suitable decision. Also, this technique requires a causal model of reality that is not always given. So, its application in autonomous navigation in the case of dynamic environments can be unprofitable.
