Direct-drive Robot Research Articles

Digital Learning Control of Servo Systems Using Neural Networks : Control System Design for Rapid Learning and Its Application

Exact inverse systems are not always obtained by using neural networks in many applications because of the output oscillation of the inverse systems. There are several sets of input signals to the neural networks, which give approximate inverse systems. In this paper, comparisons of the learning process are first made among them by a computer simulation with a parallelogram-type robot manipulator of two degrees of freedom. It is shown that the choice of the input signals has considerable influence on its learning speed. Secondly, two learning control algorithms are proposed for applying the results of a one-layer linear neural network to the control of a direct-drive (DD) robot. The algorithms are based on preliminary learning by a one-layer linear neural network, and can be used for shortening the time of on-line learning. The effectiveness of these algorithms is demonstrated by the experiment on the control of a DD robot.

Stability and control of robotic manipulators during contact/noncontact task transition

A control methodology that addresses the problem of control of robotic manipulators during a general class of task that requires the manipulator to make a transition from noncontact motion to contact motion and contact motion to noncontact motion is proposed. During noncontact motion, a control suitable for the noncontact phase of motion is applied; during contact, another control, suitable for contact motion, is applied. These different control schemes are applied to the manipulator in such a way that the overall control is discontinuous in nature. The following closed-loop behavior is achieved: (1) the closed-loop system exhibits global asymptotic stability; (2) asymptotic trajectory tracking of generalized force and position inputs is achieved; and, significantly, (3) upon inadvertent loss of contact by the manipulator, contact is reestablished and generalized forces and positions are again achieved asymptotically. Experimental results, performed on a two-degree-of-freedom direct-drive robot, support the theoretical claims.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Control of Robotic Manipulators During General Task Execution: A Discontinuous Control Approach

In this article, a control methodology is proposed that ad dresses the problem of control of robotic manipulators during a general class of task that requires the manipulator to make a transition from (I) noncontact motion to contact motion and (2) contact motion to noncontact motion. Specifically, the problem of control of a general n-degree-of-freedom rigid-link robotic manipulator during a general class of tasks, as described ear lier, is treated. It is assumed that during the contact phase of the assumed task, frictionless point contact is made with fixed objects in the manipulator work environment, which is modeled as a linear mechanical impedance. Furthermore, it is assumed that exact knowledge of the manipulator and work environment kinematic and dynamic parameters is available. The following closed-loop behavior is achieved with the pro posed control law: (1) the closed-loop system exhibits global asymptotic stability; (2) asymptotic trajectory tracking of gen eralized force and position inputs is achieved; and significantly, (3) on inadvertent loss of contact by the manipulator, contact is reestablished, and generalized forces and positions are again asymptotically achieved. This closed-loop behavior is achieved in the presence of collisions between the manipulator and ob jects in the manipulator work environment; such collisions may include multiple impacts as a result of bouncing on initial contact. A general mathematical framework is established to prove that the closed-loop robotic system, with a discontin uous control applied, is asymptotically stable. Experimental results performed on a two-degree-of-freedom direct-drive robot support the theoretical claims made in this work.

Learning hybrid force and position control of robot manipulators

The learning control is applied to hybrid force and position control of robot manipulators. When the geometry and position of a constraint surface is known, the hybrid force and position controller and the feedforward compensator can be designed in the constraint coordinates. When the operation is periodic, the learning hybrid force and position control enhance the control performance as the feedforward compensator is updated in each cycle by the force and position error in the preceding trials. This scheme is proved to be asymptotically stable. A two degree of freedom SCARA-type direct-drive robot manipulator is used to test the learning hybrid force and position control. The deburring tool mounted on the upper link of the robot could follow a flat, tilted flat, and curved 1/4 aluminum plate with a desired contact force of 10 N (within the root-mean-square force error of 1.95 N) and with a desired tangential velocity. The experiments confirmed the effectiveness of the learning hybrid force and position controller. >

Control of a Direct-Drive Arm Using Novel Repetitive Control.

A direct-drive (DD) arm system is one of the most attractive and interesting research subjects in the field of robot engineering. Up till now, many DD robot arm systems with vertical axis have been adopted for industrial applications, but very few systems with horizontal axis because of the difficulty to reduce the speed fluctuation caused by the arm weight. The PID control without any additional control process has very poor response characteristics because of the continuous load fluctuation and the time delay in the system, The accuracy is consequently, not satisfactory. Therefore, we proposed a repetitive control with learning process (position-memory type) to drive the horizontal axis robot arm with constant speed, which is based on the periodicity of the arm action.We apply this control method to the position control in which the arm reciprocates between arbitrary two points. We propose a novel control strategy in which time-memory type learning process is introduced in combination with the position-memory type learning process in order to improve the positioning accuracy. We also discuss about self tuning of the compensation rate in the system.The control method is very simple and powerful. The availability of this repetitive control method is compared with experiments.

Implementation of robust impedance and force control

In this paper, we discuss the problem of implementing impedance control in the presence of model uncertainties and its application to robot force control. We first propose a sliding mode-based impedance controller. The implementation of the targeted impedance, and the preservation of stability in the presence of model uncertainties, are the key issues in the proposed approach. Using sliding mode control, a simple and robust algorithm is obtained so that the targeted impedance can be accurately implemented without the exact model of the robot. The controller is designed in terms of the task space coordinates. The chattering in the sliding mode control is eliminated by using a continuous function. The problem of force control is also addressed for the impedance controlled robot. An off-line estimation method of the environment model is suggested and used in the force control scheme. The proposed impedance and force control schemes have been experimentally verified on a two degree-of-freedom direct-drive robot arm. The experimental results are presented in this paper.

Transferring Manipulative Skills to Robots: Representation and Acquisition of Tool Manipulative Skills Using a Process Dynamics Model

A new method based on task process models for acquiring manipulative skills from human experts is presented. In performing manipulative tasks such as deburring, a human expert moves a tool at an optimal feedrate and cutting force as well as with an appropriate compliance for holding the tool. An experienced worker can select the correct strategy for performing a task and change it dynamically in accordance with the task process state. In this paper, the human expertise for selecting a task strategy that accords with the process characteristics is modeled as an associative mapping, and represented and generated by using a neural network. First, the control strategy for manipulating a tool is described in terms of feedforward inputs and tool holding dynamics. The parameters and variables representing the control strategy are then identified by using teaching data taken from demonstrations by an expert. The task process is also modeled and characterized by a set of parameters, which are identified by using this same teaching data. Combining the two sets of identified parameters, we can derive an associative mapping from the task process characteristics to the task strategy parameters. The consistency of the mapping and the transferability of human skills are analyzed by using Lipschitz’s condition. The method is applied to deburring, and implemented on a direct-drive robot. It is shown that the robot is able to associate a correct control strategy with process characteristics in a manner similar to that of the human expert.

A comparison of model-based path control algorithms for direct-drive SCARA robots

Various advanced control strategies are applied to a direct-drive SCARA robot and studied in computer simulations. Besides computed torque control and direct adaptive control, heuristic optimal control, a new path control scheme for robotic manipulators, is included in the comparison study. PD control, the traditional robot control method, is used for generating a comparing baseline. While all schemes are applied for the same tracking task, the effect of modelling errors and measurement noise is considered in robot performance evaluation. Simulation results show that (1) without model errors, all advanced control schemes can achieve higher tracking accuracy than PD control; (2) with a random measurement error of 1%, computed torque and direct adaptive control methods are inferior to PD control; (3) heuristic control proves to be the most robust control scheme in case of mixed model and measurement errors.

Cooperative Control of Two Direct-Drive Robots Using Neural Networks. Grasping and Movement of an Object.

A learning control algorithm using neural networks is proposed for the grasping and the movement of an object by a pair of direct-drive (DD) robots of two degrees of freedom. The proposed algorithm has three feedback controllers and two neural networks. After the completion of learning, the outputs of the feedback controllers are nearly equal to zero, and the two neural networks play an important role in the control system. Therefore, the optimum setting of control parameters is unnecessary. In other words, the proposed algorithm does not necessitate any knowledge of the controlled system in advance. The effectiveness of the proposed algorithm is demonstrated by the experiment on the cooperative control of the parallelogram-type DD robots. It is also shown that gravity can be compensated by this algorithm.

Digital Learning Control of Servo Systems Using Neural Networks. Control System Design for Rapid Learning and Its Application.

Exact inverse systems are not always expected for neural networks in many applications because of the output oscillation of the inverse systems. There are several sets of input signals to the neural networks, which give approximate inverse systems. In this paper, the comparisons of the learning process are first made among them by a computer simulation with a parallelogram-type robot manipulator of two degrees of freedom. It is shown that the choice of the input signals has considerable influence on its learning speed. Secondly, two learning control algorithms are proposed for applying the results of a one-layer linear neural network to the control of a direct-drive (DD) robot. The algorithms are based on preliminary learning by a one-layer linear neural network, and can be used for shortening the time of on-line learning. The effectiveness of these algorithms is demonstrated by the experiment on the control of a DD robot.

Parameter identification of a direct-drive robot by a disturbance observer

This paper presents a dynamics identification algorithm based on a disturbance observer for a multi-degrees-of-freedom robot. In the conventional approach, the disturbance observer is used to increase the robustness of the robot. In this case, the estimated disturbance torque is directly fed back to cancel out the disturbance torque imposed on the robot. In this paper, the estimated disturbance torque is analysed from the physical point of view and the robot dynamics are identified by resolving the estimated disturbance torque. First, the basic structure of the disturbance observer is simply introduced from the motion equation of the robot. Second, the parameter identification algorithm based on the disturbance observer is explained. Finally, we show the numerical and experimental results to confirm the validity of the proposed parameter identification algorithm.

A variable structure robot control algorithm with an observer

An algorithm is proposed that combines variable structure control and observation for robot control using only joint position measurements. This algorithm is used to investigate the interactions between the observer and the controller caused by system parameter uncertainties. The theorem underlying the present work gives sufficient conditions to ensure the stability of a closed-loop system composed of a robot, an observer, and a controller. This theorem shows that the joint velocity estimation errors are bounded and the joint position tracking errors are also bounded when time tends to infinity. The experimental results achieved with the direct-drive robot DDR-1 verify the validity of the theoretical results. The experiments also demonstrate that when the end-effector of the robot tracks a 0.5-m-diameter circle at a speed of 0.5 m/s or tracks a 1-m straight line at a speed of 1 m/s, the maximum joint tracking errors are less than 0.05 degrees and the maximum trajectory tracking error is less than 0.5 mm. >

Manufacturing of a Scara Type Direct-Drive Robot with Graphite Fiber Epoxy Composite Material

SUMMARYA SCARA type direct-drive robot which can be used in the assembly operation of printed circuit boards was designed and manufactured with graphite fiber epoxy composite material. The optimum stacking angle of the graphite fiber which gives the minimum deflection of the composite robot arm was calculated by the energy method and the manufacturing method of the box-type cross section of the robot arm with composite materials was developed.In order to compare the dynamic characteristics of the composite material robot arm with a conventional material robot arm, an aluminum robot arm which has the same dimensions and thickness of the composite robot arm was manufactured.The static and dynamic characteristics of the composite material robot arm such as static deflection, natural frequency and damping as well as weight saving were considerably improved compared to those of the aluminum robot arm.

Control of two-degree-of-freedom direct-drive robot with errors of velocity measurement.

For the direct-drive robot, many excellent characteristics such as quick operation, high positioning accuracy and an application to force control are expected, but shortage of mechanical damping, difficulties of decoupling, effects of centrifugal force, Coriolis force and change of inertia become serious problems. These problems will be solved by using velocity feedback and a linearizing compensator on the premise that it is possible to obtain the velocity of the robot arm accurately. However, it is difficult to measure the velocity of the direct-drive robot accurately as compared with other types of robots. With these points as a background, this paper proposes a design method of a two-degree-of-freedom direct-drive robot control system by which the required accuracy of velocity measurement can be determined for positioning accuracy from the view point of sensitivity and robust stability. The validity of this design method is confirmed by experiments with a direct-drive robot.

Design of a load desensitized control system based on the inertial parameter root locus for a direct-drive robot

Direct-drive robots have excellent features including no backlash, low friction, and high mechanical stiffness. However, dynamic coupling between joints as well as non-linear effects become more prominent than in traditional robots with reducers. Another critical issue is that the robot becomes more sensitive to the change of load. In this paper, we develop a design method for a control system for reducing the influence of dynamic coupling and load sensitivity on direct-drive robots. For the design of the control system, we use a root locus that takes inertia as a parameter. A positive current feedback scheme is proposed in the implementation. A problem in the application of the method to a brushless DC motor is also considered. Finally, the method is implemented on a 2-DOF planar direct-drive robot with brushless DC motors as actuators. Then the validity of the method is demonstrated through experiments.

An approach to point-to-point control for direct-drive SCARA-type robots.

In recent years, attention has been paid to direct-drive for industrial robots which are essential to factory automation. Direct-drive robots, which are directly actuated with no gear reducer, have no backlash, low friction and good mechanical stiffness. This results in the excellent repeatability of positioning, that is, an order of magnitude better than that of gear-drive robots. It is, however, more prominent in the design of control system for direct-drive robots that the arm inertia variation depending on the arm configuration and the Coriolis and centrifugal torques are directly reflected to the motor axes due to the direct coupling.In this paper, a direct-drive robot with a two degree-of-freedom is developed for planer assembly tasks. The motion of the manipulator arm is constrained within a horizontal plane, so that this type of robot is referred to as a SCARA (Selective Compliance Assembly Robot Arm)-type robot. A balanced manipulator arm, the purpose of which is to cancel the inertia variation and nonlinear effects, is introduced to the direct-drive robot, thus leading to improved control performance as well as simple manipulator dynamics.In addition, the combination of feedforward compensation and a new approach to positioning control is proposed to perform stable and fast positioning without overshooting.

Adaptive force control of direct drive motor for force sensor stiffness compensation.

This paper deals with an adaptive control problem of direct drive (DD) robotic manipulators with force sensor stiffness. Since DD motor has advantages to force control in robotic application, so that many works have been reported so far in control problems. However, the dynamic compensation for the sensor stiffness has little been reported so far, even though the force sensor in DD robots decreases the stiffness of the overall DD robots. In order to make use of high stiffness of DD motors itself in the servo loop, it is necessary to take into account the sensor stiffness for the high gain dynamic control. Therefore, an adaptive dynamic compensation method is introduced here in the vein of the model reference adaptive control to widen the applicability of the DD robots, because the feedback with the constant gains may cause unstable behaviors, depending on objects dynamics. The applicability of the proposed method could be verified with some experiments. Then this method is applied to nonlinear objects, such as rotary channel. The experiments show the better performance by the proposed method for the nonlinear compensation than the conventional feed-back control method. Thus, it is shown that the DD motor can have much wider applicabilities by the proposed adaptive force control.

Adaptive force control of direct drive motor for force sensor stiffness compensation

This paper deals with an adaptive control problem of direct drive (DD) robotic manipulators with force sensor stiffness. Since DD motor has advantages to force control in robotic application, so that many works have been reported so far in control problems. However, the dynamic compensation for the sensor stiffness has little been reported so far, even though the force sensor in DD robots decreases the stiffness of the overall DD robots. In order to make use of high stiffness of DD motors itself in the servo loop, it is necessary to take into account the sensor stiffness for the high gain dynamic control. Therefore, an adaptive dynamic compensation method is introduced here in the vein of the model reference adaptive control to widen the applicability of the DD robots, because the feedback with the constant gains may cause unstable behaviors, depending on objects dynamics. The applicability of the proposed method could be verified with some experiments. Then this method is applied to nonlinear objects, such as rotary channel. The experiments show the better performance by the proposed method for the nonlinear compensation than the conventional feed-back control method. Thus, it is shown that the DD motor can have much wider applicabilities by the proposed adaptive force control.

Low-torque-ripple switched reluctance motors for direct-drive robotics

A contender for direct-drive application, the switched reluctance motor, is examined. It is first described how the number of teeth on the rotor and stator can be selected to achieve an electromagnetic gearing, with the goal of eliminating the need for a mechanical gearbox. It is also discussed how the tooth arcs of the rotor and stator can be adjusted so as to reduce the torque ripple of the motor during commutation. The treatment of these subjects includes a detailed case study in which the performance of four related motors is compared using finite-element analysis with a commutation algorithm for torque control. It is discovered that, although a tradeoff exists between peak torque and torque ripple, the maximum smooth torque and minimum torque ripple can be achieved by the same direct-drive motor design.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

A Force Control of Direct-drive Robot

This paper deals with a torque-position hybrid control of direct-drive robot. The tested robot has three-degrees-of-freedom and each joint has a direct-drive AC motor with a high resolution encoder. A direct-drive of each link realizes high precision task because no transmission exists between the shaft of the motor and the link. On the contrary, the interactive forces give significant effects to the joint motion as the equivalent gear ratio is unity. It Is essential to reduce these interactive terms in realization of direct-drive robot. Here, a disturbance observer which estimates a disturbance torque is applied to a motion control system, and an acceleration-based torque controller Is realized in the force task.