Multisensorial fusion for optimal object recognition

This work describes a sensor fusion technique proposed for the recognition of pieces under an industrial environment in which sonar techniques have been applied to obtain information about the objects, and sensor fusion technology is used to perform the identification. Bayesian probability has been used as a method for information fusion from multiple sources of probabilistic information. Since the inaccuracies within the sensors and the processing steps of the individual sensor modules are similar, conditional independence of measurements from each sensor has not been considered, and Bayesian networks have been utilized for the propagation of probabilistic information. This is a solution for an application in which ultrasonic tools appear to be a powerful technique specifically directed to industrial environments, production lines and statistical control of processes