Experimental external force estimation using a non-linear observer for 6 axes flexible-joint industrial manipulators

Abstract

is an open access repository that collects the work of Arts et Métiers ParisTech researchers and makes it freely available over the web where possible. Abstract-This paper proposes a non-linear observer to estimate not only the state (position and velocity) of links but also the external forces exerted by the robot during Friction Stir Welding (FSW) processes. The difficulty of performing this process with a robot lies in its lack of rigidity. In order to ensure a better tracking performance, the data such as real positions, velocities of links and external forces are required. However, those variations are not always measured in most industrial robots. Therefore, in this study, an observer is proposed to reconstruct those necessary parameters by using only measurements of motor side. The proposed observer is carried out on a 6 DOF flexiblejoint industrial manipulator used in a FSW process.

A novel approach for simplification of industrial robot dynamic model using interval method

“…In the research work presented by Qin et al [1], [19]- [21], as a strong external force is exerted on the robot during the machining or FSW process, the natural stiffness of industrial robot is not sufficient. To compensate the manipulator deformation, a nonlinear observer is designed to estimate the robot states (q,q,q) and the external force F.…”

“…where F denotes estimated external force, and z 4 denotes observer state 4 [1]. Obviously, the inertia matrix M(q) can be replaced by the simplified one M s (q) in (18).…”

“…kinematic and dynamic model of robot with the help of the software SY MORO+ [23], [24], or other robotic techniques. Using the Newton-Euler method or the Lagrange equations, one can get the dynamic model of the robot as the following form [1]:…”

“…In order to ensure a better tracking performance, many studies about industrial robot control have been performed. Although the machining accuracy can be raised by means of observer-based control or other methods [1], it can not avoid to bring a more complicated control structure. As a result, a great deal of calculation and memory occupation will be carried out in robot controller, affecting the real-time capability.…”

“…In the model-based control design, simple models are highly preferred. By applying the resulting simplified model to practical control system, the computation time and memory occupation will be greatly reduced, particularly in the observer-based control presented by Qin et al [1], [19]- [21]. What is more important is that this simplification is suitable for arbitrary trajectories in whole robot workspace.…”