K-FLIX—a framework for custom-configurable embedded Kalman filter microcomputers

Abstract

Kalman filter is known as one of the best linear estimators. Its use is often limited owing to the fact that computation of the next state requires considerable time or resources limiting its application scope. Recently, a new approach to Kalman filter implementation using field programmable logic has been proposed. Combining a custom-configurable microcomputer core with a Kalman filter functional unit leads to a high-speed custom-configurable low-cost Kalman filter microcomputer. The approach provides flexibility of implementation of an embedded Kalman filter with various degrees of emphasis on hardware or software depending on the system requirements in time-resource design space. This allows new applications in prediction, control and other areas in which it has not been feasible to apply it. The Kalman filter microcomputer is described and its customization discussed.