Vehicle localization systems: Towards low-cost architectures

Abstract

INS-GPS integration is a fundamental task used to enhance the accuracy of an inertial navigation system alone. However, its implementation complexity has been a challenge to most embedded systems. This paper proposes a low-cost FPGA-based INS-GPS integration system, which consists of a Kalman filter and a soft processor. Moreover, we also evaluate the navigation algorithm on a low-cost ARM processor. Processing times and localization accuracy are compared in both cases for single and double precision oating-point format. Experimental results show the advantages of the FPGA-based approach over the ARM-based approach. The proposed architecture can operate at 100 Hz and demonstrates the advantage of using FPGAs to design low-cost INS-GPS localization systems.