Dynamic localisation of autonomous guided vehicles

As autonomous guided vehicles become increasingly established within a diverse range of applications, the need for efficient and flexible operation becomes apparent. Existing localisation techniques have tended to offer either “move look update” operation, navigating from an a priori map, or continuous operation dependent upon artificial beacons placed within the environment. We describe an extended Kalman filter based navigation system which maintains a robust position estimate throughout continuous operation. In order to achieve dynamic operation, we exploit the recursive nature of the Kalman filter and utilise the higher data acquisition rate offered by infra red scanning to obtain observations of the operational environment. In order to further enhance performance, the a priori map is segmented into an array of regions containing sub-lists of features which can be rapidly matched with an observation, thus minimising the computation overhead due to data association