A MULTICRITERIA APPOACH FOR OPTIMAL TRAJECTORIES IN DYNAMIC PARAMETER IDENTIFICATION OF PARALLEL ROBOTS Xabier Iriarte, Miguel Díaz-Rodríguez, Vicente Mata Departamento de Ingeniería Mecánica, Energética y de Materiales, Universidad Pública de Navarra, 31006, Pamplona, España e-mail: xabier.iriarte@unavarra.es Departamento de Tecnología y Diseño Universidad de Los Andes, Merida, Venezuela. e-mail: dmiguel@ula.ve Departamento de Mecánica y Materiales, Universidad Politécnica de Valencia, Camino de Vera, 46020, Valencia, España e-mail: vmata@mcm.upvnet.upv.es ABSTRACT- When dealing with the development of accurate robot dynamic models, the estimation and validation of the dynamic parameters through experiments becomes necessary. One of the most important objectives to achieve in the design of the experiment, aimed at identification, is to properly excite the system so that the unknown parameters can be accurately estimated. It is customary to find proper excitations optimizing the observation matrix of the model w.r.t. a certain criterion. In this paper the suitability of some trajectory optimization criteria are evaluated for parallel robots. Moreover, a multicriteria algorithm is proposed in order to reduce the deficiencies derived from the single criterion optimization. KEYWORDS: Robot, dynamics, identification, dynamic parameters, multicriteria optimization. INTRODUCTION The need for increasing the speed of robotic systems, without loss of accuracy at the end effector robot positioning, requires the implementation of model-based control algorithms, see for instance [1]. The accuracy of the models depends on the values assigned to the parameters of the model. Geometrical parameters are usually well known through CAD models or direct measuring. However, the dynamic parameters (inertial and frictional) are not likely to be estimated through direct measuring or CAD models. Disassembling the robot is not practical and frictional behaviour depends on the operational conditions. These drawbacks make the experimental dynamic parameter estimation necessary in order to obtain