On the other hand, the second family of methods, let’s say the data- based methods, for addressing such vibration control problems has re- ceived very limited attention in robotics. An interesting c
On the other hand, the second family of methods, let’s say the data- based methods, for addressing such vibration control problems has re- ceived very limited attention in robotics. An interesting control system that consists of a fuzzy and a typical Proportional (P) controller has been employed in [6]. The controller uses force and torque measurements from the robot and achieves the effective suppression of high frequency vibration at the free-end of a stainless steel ruler. Yet, once the controller is activated the force at the robot’s wrist increases suddenly and becomes very strong for some seconds before the vibration at the ruler’s free-end is attenuated. This may lead to permanent deformation of the manipu- lated object or to undesired damage in the case of more brittle objects. In addition, as in most cases with fuzzy controllers [7] the fuzzy rules, the deffuzification as well as other parameters should be appropriately manually re-tuned for different flexible objects and robots constituting thus a time-consuming and relatively involved procedure. The same au- thors presented in [8] a method for the suppression of the residual vi- bration of a flexible beam through a template matching procedure based on which the free vibration response of the beam’s free-end is appropri- ately matched with a typical sine-type function. This method assumes that the flexible object behaves like a linear spring with constant stiff- ness and necessitates a delicate procedure for the determination of the type and frequency of the free vibration as well as for the characteristics of the adjustment-motion strategy that leads to the attenuation of the beam’s free-end vibration.
In general, the vibration control problem of inpidual cantilever flexible beams has been addressed in many studies (see [13–15] and references therein) based on physical or data-based models and addi- tional sensors and/or actuators applying the control input at the tip of the beam. Yet, this is not the case when a flexible beam is manipulated by a robot in an industrial line, where the use of additional equipment on the beam and/or the robot is not desirable. More related vibration control problems have been studied for flexible robotic links [9,10] and anti-sway control systems [11,12] using physical models for the investi- gated robots and cranes in appropriate Input-Shaping-Techniques (IST). In summary, an industrial robot that is used in a typical production line may manipulate various flexible objects characterized by different physical parameters some of which are unknown or difficult and time consuming to be determined. The dynamics of the manipulated flexi- ble object affects the robot dynamics and cannot be neglected in the design of a vibration control system for suppressing the residual vibra- tion at the free-end of the flexible object. On the other hand, the use of additional visual equipment, sensors, active damping systems or other mechanisms in a relative control system reduces the versatility of the production line to adapt to the manipulation of different flexible objects and increases the total production cost. Furthermore, a vibration control system should be capable of operating without human intervention be- ing insensitive to various external disturbances and easily adaptable to operate with various flexible objects. Last but not least, the robot cannot be removed from normal production for long periods so as to be used for the design and fine-tuning of a control system.
The goal of the present study is the development of a novel method for the vibration control of flexible beams manipulated by industrial robots overcoming the above mentioned difficulties. The design of the present control system includes: (i) The AutoRegressive with eXogenous (ARX) input stochastic modeling of the robot-beam system based ex- clusively on experimental data provided directly from the embedded robot’s force sensor at its wrist; (ii) the design of an outer control system that is founded on the ARX model of the previous step. This system con- sists of a typical feedforward PID-type controller and a feedback that enables the attenuation of the force at the robot’s wrist and thus the suppression of the vibration at the beam’s free-end; (iii) a synthetic en- vironment within which the performance of the control system may be investigated with accuracy under different operating conditions through simulations before its final implementation in the industrial robot.