Date of final exam: 01/02/2000

E-mail: cinatale@unina.it

Tutor: Prof.  B. Siciliano, Università degli Studi di Napoli Federico II

  ___________________________________________________________________________________________________________

Six-DOF Interaction Control of Robot Manipulators   ___________________________________________________________________________________________________________

Advisor:

Prof.  B. Siciliano, Università degli Studi di Napoli Federico II

Summary of the thesis

The main topic of this dissertation is control of the interaction of robot manipulators with the environment, during the execution of six-degree-of-freedom tasks.
The work is aimed at demonstrating how the problem of controlling the rotational degrees-of-freedom cannot be solved by a mere analytical extension of the results obtained for the translational degrees-of-freedom.
In this farmework, the well-known technique of resolved acceleration control is reviewed, to the purpose of highlighting the main drawbacks of the usual representations of end-effector orientation, i.e., Euler Angles. A class of geometrically meaningful angle/axis-based descriptions of orientation is introduced as an effective way to counteract the problems caused by the adoption of minimal representations, i.e., representation singularities and lack of geometrical meaning. Among the different angle/axis representations, the unit quaternion is recognized as an efficient and singularity-free mathematical tool, which leads to globally stable tracking controllers.
The two main interaction control strategies are takled, namely ''impedance control'' and ''force Control''. A new impedance control scheme is proposed thanks to the adoption of a class of geometrically meaningful rotational displacements. The impedance equations are derived by resorting to an energy-based argument. This allows the impedance equation to satisfy the property of ''task geometric consistency''. Furhter, if among all the possible descriptions of rotational displacements, the unit quaternion is chosen, the occurrence of representation singularities is completely avoided. The control scheme is shown to be valid also for redundant manipulators, if a suitable control action is designed to stabilize the inner motion. The task space impedance control is compared with the classical operational space formulation from both a theoretical and experimental point of view.
The experiments have been performed on the two industrial robots, equipped with force/torque sensors, available in the PRISMA Lab. This has been possible thanks to the open control architecture endowing the robots. This kind of control unit allows the user to directly control motor currents.
The proposed task space force control scheme belongs to the well-known parallel force/position control strategy, and it is designed for a six-degrees-of-freedom task; hence, by taking into account the problem of task geometric consistency. Once again, the unit quaternion is used to conceive a controller which satisfies this property and is singularity-free.
The second part of the work is devoted to show the effectiveness of the proposed approach and its applicability at industrial level. To this purpose, a few complex workcell tasks are performed on a dual-robot system. Cooperatiive manipulation is considered at two different levels, namely at task planning level and at control level. In the former case, one of the two robots is position controlled and programmed in the native standard lamguage, while the other robot is controlled at current level in order to apply one of the interaction control strategies presented in the first part. The results of a peg-in-hole task and a pressure-forming task are presented, which show the real effectiveness of the proposed control algorithms. In the latter case, the robots rigidly hold a rigid object interactiong with a compliant environment. They are both controlled in the open operating mode and an object-level impedance control is accomplished, by resorting to the symmetric task space formulation.
Even though the obtained results can be considered really satisfactory, the thesis' work is not conclusive. For example, the results obtained for the cooperative manipulation problem are only preliminary and a lot of work should be done in the field of internal force control. As future research directions are indicated the interaction control of flexible manipulators, the impelmentation of the geometrically meaningful techniques to haptic interfaces and the control of complex mechanical systems, i.e., vehicle-manipulator systems.
 

  _______________________________________