A system identification technique for haptic devices

The number of unknown parameters in a high degree-of-freedom (DOF) robot makes traditional identification techniques difficult to implement. A non-conventional system identification method is proposed to model the Freedom 6S, a force feedback hand controller with 6 DOF. Compared to industrial robots, haptic devices are back-drivable, and they are characterized by relatively large forces and low friction. Our method is used to identify parameters such as the mass, centroid, inertia and friction of the manipulator. The number of unknown parameters in the dynamic equations describing the system can be reduced by physically locking down selected joints. This reduced parameter set can then be identified by a method of nonlinear optimization, using measured data as a basis for cost-function evaluation. By an appropriate sequence of joint analysis, all the parameters can then be identified. This paper discusses the systematic locking order of joints with the objective of reducing the number of parameters requiring identification at each stage. The method is verified by comparison between simulated and measured dynamic data