Development of an particle swarm algorithm for vehicle localization

Abstract

This paper describes the development of a filter algorithm based on the behaviour of biological swarms. The main goal of the algorithm is to perform vehicle localization by combining the data from different sensors - GPS, IMU, speedometers, etc. - and digital maps. In this sense, the algorithm considers several solutions at the same time like Particles Filters. The algorithm has been developed off-line using real data captured from an instrumented vehicle at LIVIC. Performance of the algorithm has been validated and compared with and EKF with encouraging results.