A Comparison of H/spl infin/ with Kalman Filtering in Vehicle State and Parameter Identification

The Kalman and H_infin filters, which aim to minimize separate criteria, are optimal only in ideal circumstances. A question that arises in practice is how to determine which filter performs better using a posterior common criterion pertinent to the application. To address this issue, the authors use the mean squared error as a measure of performance while ensuring that the filters´ tuning parameters are also comparable. Analysis of combined state variable and parameter estimation, in the area of vehicle dynamics, has shown that the H_infin filter has the ability to outperform the Kalman filter as long as the respective Riccati equations start from the same initial condition