Location Estimation using Delayed Measurements

When combining data from various sensors it is vital to acknowledge possible measurement delays. Furthermore, the sensor fusion algorithm, often a Kalman filter, should be modified in order to handle the delay. This paper examines different possibilities for handling delays and applies a new technique to a sensor fusion system for estimating the location of an Autonomous Guided Vehicle. The system fuses encoder and vision measurements in an extended Kalman filter. Results from experiments in a real environment are reported. 1 Introduction In a complex control system more than one sensor is often needed to generate and maintain a reliable state estimate. Furthermore, the complexity of the computations in sensors will often introduce significant time delays from the acquisition to the availability of a measurement. This can be due to signal and image processing, data transportation, laboratory tests, etc. Two examples of systems with delayed measurements are an Autonomous Guided Vehicle..