Statistical Approach for Bias-free Identification of a Parallel Manipulator Affected by Large Measurement Noise

The problem of high measurement noise in identification issue is treated in this paper for an innovative parallel robotic manipulator. To consider the noise and the correlation across the system’s output a complete statistical approach is presented. The Maximum-Likelihood estimator is used for the identification of the dynamics parameters. Furthermore the experiments were designed based on a statistical criterion, such that the resulting excitation trajectories minimize the uncertainty bounds of the estimation. The experimental results are consequently compared with those resulting from classic deterministic approaches. This comparison demonstrates that the presented methodology yields bias-free and asymptotic efficient estimation.