Chowarit Mitsantisuk

FPGA-Based High-Performance Force Control System With Friction-Free and Noise-Free Force Observation

In this paper, a novel force-sensing method is proposed for a high-performance force control system based on friction-free and noise-free force observation. A periodic signal is inserted into the control system for friction reduction. A combination of a high-order disturbance observer (DOB) (HDOB) and a Kalman filter is constructed to perform the force sensing. The HDOB is designed to estimate the force and reject oscillatory disturbances in the estimation. The force-sensing bandwidth is improved through effective noise suppression by the Kalman filter. Additionally, this paper proposes the application of the HDOB to the bilateral control system of a different master-slave mechanism. All the control algorithms are implemented in a field-programmable gate array with a high sampling rate that also enables the widening of the bandwidth of the force control system. The effectiveness of the proposed method is verified by experimental results.

Sinteza bezsenzornog upravljanja silom za fleksibilnog robota korištenjem upravljanja omjerom rezonancija temeljenim na metodi koeficijentnog dijagrama

Generally, the flexible robot system can be modeled as the two-mass system which consists of a motor and load connected by a spring. Thus, its elasticity causes resonance in the system. By using the conventional PID controller, this method cannot perform well in this situation. Much research has proceeded with the aim of reducing vibration. A new effective control method, the resonance ratio control, has been introduced as a new way to guarantee the robustness and suppress the oscillation during task executions for a position and force control. In this paper, three techniques are proposed for improving the performance of resonance ratio control. Firstly, a new multi encoder based disturbance observer (MEDOB) is shown to estimate the disturbance force on the load side. The proposed observer is not necessary to identify the nominal spring coefficient. Secondly, coefficient diagram method (CDM) has been applied to calculate a new gain of the force controller. A new resonance ratio gain has been presented as 2.0. Finally, the MEDOB and load side disturbance observer (LDOB) are employed to identify a spring coefficient of flexible robot system. By using the proposed identification method, it is simple to identify the spring coefficient and easy to implement in the real flexible robot system. The effectiveness of the proposed identification method is verified by simulation and experimental results.

Transparency improvement in a bilateral motion-scaling control using Kalman-filter-based disturbance observer

This paper proposes a new wideband bilateral motion-scaling control to apply in the small size robot system. The scaling Hadamard matrix is applied to reduce or magnify the force and position response of bilateral motion-scaling control. The technique to estimate the action/reaction forces of master-slave robot is designed using Kalman-filter based disturbance observer (KFDOB). Optical encoder and acceleration signals were fused to estimate contact forces occurring during an experiment. With the bilateral motion-scaling control and KFDOB, we can control the small scale master-slave robot with high precision. Comparative experiments show the KFDOB and its multi-sensor fusion to provide performance superior to that of previously disturbance observer. As seen in the experimental results, it is clear that high transparency and good perception of the environmental stiffness with the response of the master-slave robot can be achieve successfully. An additional result of FFT analysis which guarantees noise compensation is also presented.

Force sensation improvement in bilateral control of different master-slave mechanism based on high-order disturbance observer

In this work, we present a bilateral control system with different mechanisms of master and slave sides. A linear shaft motor and a ball screw perform the roles of the master and slave, respectively. To reduce the friction effect on force estimation due to the ball screw mechanism, a periodic signal is inserted into the control signal of the slave side. The addition of periodic signal causes oscillatory force responses on both master and slave sides. Therefore, a high-order disturbance observer is designed for force sensing operation on the slave side to reduce the effect of oscillatory disturbance on force information. The control algorithm consists of a conventional disturbance observer and a high-order disturbance observer for master and slave sides, respectively. The numerical simulation results verify the effectiveness of the proposed method.

Haptic human-robot collaboration system based on delta robot with gravity compensation

Human-robot collaboration systems have attracted more and more researchers attention. Many experiments in a new design of robot system illustrated that it can be used to help and support human operator. In the past few decades, delta robots have popular usage in industrial applications, especially in picking and packaging for their relative advantages such as high speed, high accuracy, and light weight devices. Delta robot is a parallel manipulator that consists of three arms connected to universal joints at the base. This paper presents work on a new force sensor less haptic human-robot collaboration system based on 3D parallel mechanism delta robot. The controller of delta robot was designed by using force sensorless bilateral control approach with considering of gravity compensation. Each joint of the AC servo motor can be programmed to estimate the external force from unknown environment by using the disuturbance observer (DOB). It was shown that delta robot can not only be applied for industrial applications, it can also be conveniently used for human-robot interaction. Experimental results are presented which demonstrate that the gravity effect can be compensated. Therefore, our proposed robot system is suitable for human-robot collaboration system.

Parameter estimation of flexible robot using multi-encoder based on disturbance observer

Generally, the flexible robot system can be modeled as the two-mass system which consists of a motor and load connected by a spring coefficient. In order to achieve accurate tracking objectives, the correct of parameter identification in the model based controller is necessary. The objective of this paper is the parameter identification of a spring coefficient of flexible robot system. The multi-encoder based disturbance observer (MEDOB) and load side disturbance observer (LDOB) are employed to identify a correct spring coefficient parameter of flexible robot system. The comparison of external force estimation between MEDOB and LDOB is used to find the correct value of spring coefficient. By using the proposed identification method, it is simple to identify the spring coefficient and easy to implement in the real flexible robot system. The effectiveness of the proposed identification method is verified by simulation and experimental results.

Robotics-assisted rehabilitation therapy for the hands and wrists using force sensorless bilateral control with shadow and mirror mode

Recently, the robotic devices are used mainly as therapy aids in the rehabilitation process of a disabled persons. The robotic device are designed with applications of techniques to support the adaptive exercise possible. It can apply to repeat the process exercise as many times as you wish. Moreover, the data measurements of their improvement or decline can be analyzed through the sensors on the device. This paper proposes a specially designed force sensorless bilateral control based on Kalman-Filter Based State Observer (KFSO) for the hand and the wrist of the patient. By using KFSO, it can be applied to estimate a high bandwidth force sensation and good perception of an unknown environmental stiffness. Moreover, bilateral control based on the shadow mode and mirror mode is also proposed in order to improve human hand movement. To successfully implement neuro-recovery process, the mirror mode is proposed to help the brain into thinking the weaker hand or wrist is moving according to the stronger hand or wrist. With such a robust bilateral control scheme, significant high bandwidth haptic sensing in a rehabilitation robot can be achieved. The simulation results of the rehabilitation robot are provided to illustrate the performance of the proposed algorithms.

Force sensorless control with 3D workspace analysis for haptic devices based on delta robot

Recently, the human-robot interaction systems are extending their capability to work in many applications. Appropriate haptic devices or robot systems have an important role to support the collaboration task which require close interaction between human and robot. This paper proposes a novel 3D parallel mechanism delta robot based on Kalman filter based state observer for the haptic devices. In the mechanical design, a 3D parallel mechanism delta robot contains three identical and symmetrical upper-arm and fore-arm of parallel links that are connected to the base plate with motor at lower end and to the end-effector with a ball joint at top end. The delta robot are designed and analyzed in order to realize a 3D motion in the workspace of the robots end-effector. In the controller design, Kalman filter based state observer, (KFSO) can be applied to estimate an external torque of motor. With such a force sensorless control scheme based on the feedforward disturbance torque compensation loop, significant high bandwidth haptic sensing in a delta robot system can be achieved. The proposed method has been successfully applied in a real delta robot system in order to show its applicability. The experimental results of the rehabilitation robot are provided to illustrate the performance of the proposed algorithms. In the experimental results, the system was able to estimate a 3D motion and external force in the workspace of the robots end-effector.

Bilateral Control Based on Disturbance Observer of Delta Robot with Gravity Compensation

Recently, delta robots is widely used in many industrial application because the structure of delta robot is close loop chain and have properties of high rigid body and fast movement. However, the moving part of delta robot is arranged in pattern that influenced by gravity and has an impact on control system. In this paper, we proposed force and position control in master-slave configuration while compensate gravity. The control system using disturbance observer (DOB) to estimate sum of action/reaction force and gravity force of master-slave delta robot instead of the force sensors. The external force on travelling plate of delta robot is estimated by disturbance force on upper and computational gravity force from dynamic model. Moreover, the delta robot may various in size due to the conditioning of workspace, therefore the method to control the task space is proposed. The simulation is conduct to verify the control system, compare the position and force of travelling plate with and without gravity compensation.

Multi-sensor fusion observer based multilatral control of haptic devices without force sensor

Among all the sensor fusion approaches, Kalman filter is the most widely used technique for system state estimation. The Kalman filter has many uses, including applications in tracking objects, navigation, economics, computer vision and robot controlling. As an example application, it is suitable method for compensating noise measurements from sensor data input and also for sensor fusion. In this paper, a novel multilateral control based on observer technique has been proposed by using disturbance observer with Kalman filter, named Kalman-Filter based Disturbance Observer (KFDOB). Kalman-filter is designed to estimate velocity response and disturbance observer is used to estimate the action/reaction force of haptic devices. The design of multilateral control based on the Hadamard matrix is also presented. With such a robust sensorless force control scheme, high bandwidth of force sensing in multi-robot system may be achieved since acceleration sensor and optical encoder are used to estimate the external force information. Moreover, its measured noise is also significantly reduced. The experimental results are provided to illustrate the performance of the proposed algorithms.