Sequential identification of base parameters of robots

A common idea to identify the inertial parameters of robots is to use linear model in the parameters and standard least squares (LS) techniques. These techniques are sensitive to noise measurements and error modelling and need to use exciting trajectories which are difficult to generate when there is a large number of parameters to identify. To overcome this difficulty, the paper presents a sequential method using an energy model to identify the base inertial parameters of a 3 degrees of freedom robot. Each test motion uses an exciting trajectory for the moving joints and an exciting configuration for the fixed ones, and allows the identification of a small number of parameters. Bias and variance of the LS estimator are studied and compared to those of the global LS solution, using analytical results or statistical simulation techniques.<>