M.N. Sluzhivyi, D.A. Gurman

An analysis of autonomous moving vehicle (AMV) position estimation algorithms effectiveness is of great practical interest when studying moving vehicle navigation systems, operating in the conditions of many interfering factors influencing AMV position estimation accuracy. In the work inertial-sensor-based and terrain-landmark-based (SLAM algorithm) data AMV position estimation algorithms are studied. Acceleration measurement accuracy depends on the errors of inertial sensors whereas SLAM algorithm errors depend only on landmark (object found on the terrain, e.g. by sonar) range and bearing measurement accuracy. Two source data fusion application which ensures AMV coordinates determination accuracy enhancement is considered. Integrated data processing is very efficient at inertial system failure or at absence of a near-located landmark on the terrain as well as in case of abrupt error increase of one of the sources. It is shown that AMV acceleration correlation coefficient increase and acceleration variance increase with fixed correlation cause AMV coordinates estimation error variance increase. The results of AMV position determination error estimation at various AMV accelerations and landmark coordinates determination errors with application of two-source data fusion on the basis of weight processing are presented as well. All the results were obtained at normally distributed measurement noises which is quite suitable for the majority of practical cases.