Methods and Algorithms for Sensor Data Fusion Aimed at Improving the Autonomy of a Mobile Robot

A basic requirement for an autonomous mobile robot is to localize itself with repect to a given coordinate system. In this regard two different operating conditions exist: structured and unstructured environment. The relative method and algorithms are strongly influenced by the a priori knowledge on the environment where the robot operates. If the environment is known, a proper multisensor system endowed with an efficient data fusion algorithm may provide a very accurate localization. In this chapter the localization problem is formulated in a stochastic setting and a Kalman filtering approach is proposed for the integration of odometric, gyrocope, sonar and video camera measures. If the environment is only partially known the localization algorithm need preliminary definition of suitable environment map. Different probabilistic methods for sensory data fusion aimed at increasing the environment knowledge are proposed and discussed.