“In a different direction, the necessity to model the analysis of noisy incomplete sensory data not by logic but by Bayesian inference first came to the forefront in the robotics community with their use of Kalman filters.”