Review of modelling and control of flexible-link manipulators

Abstract

In recent decades, flexible manipulators have been studied by many researchers from robotics, solid mechanics, and control fields. Flexible manipulators have many advantages, including low weight because of the slenderness of the links of the robot. Although the original objective was to take advantages of the slenderness or flexibility in real robots, the challenging dynamics of the systems intrigued interests to employ an experimental flexible manipulator as a test bed to evaluate different modelling or control methods. With such a vast and various literatures, a review is indispensable for researchers who want to adapt their interests with the area. Some valuable review articles have been published, referencing numerous articles on single-link or multi-link flexible arms. This article pays an inclusive focus on trends of the research on modelling and control of multi-link flexible-link manipulators. The scope of this review article is particularly on two-link flexible manipulators, relevant models presented for closed-loop applications, and model-based control. Recent and historical contributions in the modelling and control of flexible-link manipulators are presented and discussed. As regular industrial manipulators normally have multiple links with two long links, that is, upper arm and forearm, this review can introduce advances in considering elasticity effects to robotic researchers.

Dynamic modeling and adaptive robust boundary control of a flexible robotic arm with 2‐dimensional rigid body rotation

“…Remark 2. The adaptation law (17) yields positive values foṙk 3 ,̇k 3 , anḋk 2 at all times. Accordingly, the dynamic control gains will be continuously increasing by time and reach high values.…”

“…Accordingly, in order to avoid reductions in safety, reliability, and accuracy associated with undesirable vibrations, many researchers have put the study of flexible manipulators on the agenda. [1][2][3][4] Studies revolve around dynamic modeling, control design, and sensor/actuator systems. In dynamic modeling, the challenge mostly lies in the incorporation of flexibility effects in the system model, leading to the increase in complexity that, in turn, complicates the problem of controller synthesis.…”

“…The main techniques used for model truncation in flexible manipulators are the lumped-parameter, assumed-mode, and finite element methods. 3 In this regard, a truncated model of the PDE system is obtained by considering only a finite number of vibration modes for control design and ignoring the remaining high-frequency modes known as residual modes. However, a finite-state approximation to a PDE may wrongly render fundamental system theoretic properties like controllability and observability into functions of the approximation.…”

“…For example, there has been a tendency toward using distributed piezoelectric actuators and sensors for active vibration control in flexible manipulators, but they require high voltage and expensive amplifier and interfacing circuits. 3 Therefore, to achieve economical and realistic control systems and to overcome the hardware complexities of distributed control approaches, boundary control by stabilizing the flexible system only through its boundary without intruding into the domain sounds promising. 8,9 This has recently motivated different studies on the PDE-based boundary control of the flexible structures such as string, 10 rod, 11 beam, [12][13][14] plate, 15,16 marine riser, 17 satellite, 18 rotor, 19 wind turbine tower, 20 systems subject to moving mass, 21 moving belt, 22,23 and gantry crane.…”

Reconfigurable control of flexible joint robot with actuator fault and uncertainty

“…In 1987, Spong proposed in [2] a simplified model for the flexible articulation of the robot. After this, a large number of theoretical and experimental investigations were carried out on the control of robots with flexible joints: Singular perturbation and integral variety, feedback linearization, cascade system and integral backstepping control, PD control, adaptive control, robust control, neural network control, fuzzy control and some other commands [3][4].…”