Ph.D. received on: 8/7/1997

E-mail: rfm@ipa.fhg.de

Tutor: Prof.  Alessandro De Luca, Università degli Studi di Roma “La Sapienza”
 
 

 ___________________________________________________________________________________________________________

A general methodology for task-based modeling and control of cooperating robotic systems   ___________________________________________________________________________________________________________

Advisor:

Prof. Alessandro De Luca, Università degli Studi di Roma “La Sapienza”

Summary:

This dissertation discusses the modeling and control problems for a robotic cooperating system, i.e., a system constituted by two or more manipulators in contact with a common payload. The most in-teresting issues related to this class of systems are the interaction of each robot with a dynamic envi-ronment, and the presence of kinematic constraints on system motion, mainly due to the contact with the common payload. The actual form of these dynamic interactions and kinematic constraints de-pends on the particular system structure, namely, on the number of manipulators, the type of contact between each of them and the payload, the kinematic/dynamic behavior of the payload itself, the pos-sible presence of further external constraints on system motion.
In any resulting situation, a different set of variables can be chosen to characterize the kinematic and dynamic behavior of the overall system. This choice, that constitutes the first, fundamental step of dynamics modeling, is obviously not unique. A natural requirement, however, is that the system task can be easily described by the same set of parameters used for modeling its dynamics. Furthermore, the chosen description should allow studying the system controllability and, in particular, the possi-bility of assigning the desired outputs’ behavior in an independent way.
The proposed methodology can be applied to a generic cooperating system (in terms of number, type and mobility of the manipulators, payload grasp, and kind of payload), and provides a systematic, task- and control-oriented way to choose motion and force parameters for dynamics modeling. Fur-thermore, it allows individuating proper sets of outputs that make possible to decouple and linearize input-output the resulting system model. In particular, for any chosen parametrization, two equiva-lent sets of outputs can be found and, correspondingly, two equivalent dynamic models, that can be input-output decoupled and linearized. In fact, in the task-space of a cooperating system, generalized directions exist, where an energy exchange between robots and environment occurs. Along these di-rections, the system behavior can be described by motion or force parameters in an equivalent way, giving rise to two alternative dynamic models.
On the resulting input-output decoupled and linearized system, any suitable control law (e.g., a stan-dard linear one) can be finally applied to get the desired behavior for the chosen output variables.
 

 _______________________________________