FPGA implementation of the EKF algorithm for localization in mobile robotics using a unified hardware module approach

In this paper a Hardware Architecture for computing the Extended Kalman Filter (EKF) is presented which is addressed to solve the self-localization problem of autonomous mobile robots. In this case, the overall EKF algorithm has been implemented in hardware over an Altera Cyclone IV FPGA with a Nios II processor, in which the latter is used only for interfacing and communication tasks. The achieved implementation has been adapted and applied to the mobile platform Pioneer 3AT (P3AT) for validation task. The prediction stage of the EKF algorithm was based on a dead-reckoning system model and the estimation stage on a measurement system which uses a sensor of Laser Range Finder (LRF). The proposed architecture has been designed considering a Unified Hardware Module approach using floating-point arithmetic operators, allowing the operations to be computed with large precision and dynamic range. Furthermore, several metrics have been used to evaluate the system performance, measuring both FPGA resources consumption, power consumption and execution time. Finally, the suitability of reconfigurable devices for such kind of applications has been verified and also discussed.