Abstract—This paper describes a Hardware/Software
Co-design approach for the Extended Kalman Filter (EKF)
applied to the localization problem in mobile robotics. The EKF
algorithm has been implemented and run on an Altera Cyclone
IV FPGA with a Nios II embedded processor jointly with
specific hardware modules, being adapted and applied to the
mobile platform Pioneer 3AT (P3AT). In order to achieve this,
we developed both the model of the mobile robot and its
measurement systems previously to obtain the respective EKF
equations. In the prediction step of the EKF algorithm, a system
model based on concepts of dead-reckoning has been used and
its implementation was achieved in software, using the Nios II
processor. Conversely, in the estimation step of the EKF
algorithm the respective equations have been implemented
directly in hardware, producing an overall balanced
implementation.
Index Terms—Dead-reckoning, extended Kalman filter,
FPGA, hardware/software co-design, Pioneer 3AT.
L. Contreras, S. Cruz, C. H. Llanos, and J. M. S. T. Motta are with the
Graduate Program of Mechatronics Systems, Department of Mechanical
Engineering, University of Brasilia – UnB, Brasilia, Brazil (e-mail:
lcontreras@aluno.unb.br, sergio@unb.br, jmmotta@unb.br, and
llanos@unb.br).
Cite: Luis Contreras, Sérgio Cruz, J. M. S. T. Motta, and Carlos H. Llanos, "Hardware and Software Co-design for the EKF Applied to the Mobile Robotics Localization Problem," International Journal of Machine Learning and Computing vol. 5, no. 2, pp. 101-105, 2015.