• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.28B no.9
• /
• pp.739-746
• /
• 1991

This paper develops a modified multilayer perceptron architecture which speeds up learning as well as the net's mapping accuracy. In Phase I, a cluster partitioning algorithm like the Kohonen's self-organizing feature map or the leader clustering algorithm is used as the front end that determines the cluster to which the input data belongs. In Phase II, this cluster selects a subset of the hidden layer nodes that combines the input and outputs nodes into a subnet of the full scale backpropagation network. The proposed net has been applied to two mapping problems, one rather smooth and the other highly nonlinear. Namely, the inverse kinematic problem for a 3-link robot manipulator and the 5-bit parity mapping have been chosen as examples. The results demonstrate the proposed net's superior accuracy and convergence properties over the original backpropagation network or its existing improvement techniques.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2002.12a
• /
• pp.80-83
• /
• 2002

정밀한 서보기술에 바탕을 두고 있는 산업용 로봇 팔은 기계적인 강성도와 서보기구들로 인하여 정확한 제어가 가능하다 그러나 로봇팔은 동작중비 관성모멘트의 변화와 기하학적인 구조로 인하여 비선형적인 시스템이다. 특히 관성모멘트의 변화는 로봇팔의 회전속도와 로봇팔이 이동시키는 대상물의 무게 등에 따라 관성모멘트는 변할 수 있다. 이로 인하여 관성모멘트의 변화에 따라 로봇팔의 정확한 위치제어가 힘들어진다. 따라서 본 논문에서는 관성모멘트의 변화에도 불구하고 강인한 응답특성을 나타내는 제어기의 설계가 요구된다. 일반적으로 PID .제어기는 설계의 용이성으로 산업현장에서 널리 사용된다. 그러나 PIB 제어기의 각 계수값을 설정하는데 많은 어려움이 발생한다. 따라서 본 논문에서는 PID 제어기의 각 계수 값을 퍼지알고리즘을 이용하여 자동으로 설계할 수 있는 자기동조 PID 제어기를 설계한다.

• The Transactions of the Korean Institute of Electrical Engineers
• /
• v.43 no.5
• /
• pp.838-847
• /
• 1994

Although there is an analytical proof of modeling capability of the neural network, the convergency error in nonlinearity modeling is inevitable, since the steepest descent based practical larning algorithms do not guarantee the convergency of modeling error. Therefore, it is difficult to apply the neural network to control system in critical environments under an on-line learning scheme. Although the convergency of modeling error of a neural network is not guatranteed in the practical learning algorithms, the convergency, or boundedness of tracking error of the control system can be achieved if a proper feedback control law is combined with the neural network model to solve the problem of modeling error. In this paper, the neural network is introduced for compensating a system uncertainty to control a nonlinear dynamic system. And for suppressing inevitable modeling error of the neural network, an iterative neural network learning control algorithm is proposed as a virtual on-line realization of the Adaptive Variable Structure Controller. The efficiency of the proposed control scheme is verified from computer simulation on dynamics control of a 2 link robot manipulator.

• Proceedings of the KIEE Conference
• /
• 1999.07g
• /
• pp.3077-3079
• /
• 1999

A visual servoing method has been proposed based on disturbance observer to eliminate the effect of the off-diagonal component of image feature Jacobian, since performance indices such as measurement sensitivity of visual features, sensitivity of the control to noise and controllability could be improved when an image feature Jacobian was given as a block diagonal matrix. In this paper, experimental results of disturbance observer-based visual servoing are discussed where Samsung FARAMAN-ASl 6-axis industrial robot manipulator is employed. Also, the feature saturator is proposed to stabilized the disturbance observer loop by saturating the differential changes of the image features.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 1993.06a
• /
• pp.881-886
• /
• 1993

Fuzzy logic rule-based controller has many desirable advantages, which are simple to implement on the real time and need not the information of structure and dynamic characteristics of the system. Thus, nowadays, the scope of the application of the fuzzy logic controller becomes enlarged. But, if the controlled plant is a time-varying and nonlinear system, it is not easy to construct the fuzzy logic rules which usually need the knowledge of an expert. In this paper, an approach in which the logic control rules can be self-organized using genetic algorithm will be proposed and the effectiveness of the proposed method will be verified by computer simulation of the 2 d.o.f. planar robot manipulator.

• Journal of the Korean Society for Precision Engineering
• /
• v.13 no.3
• /
• pp.102-109
• /
• 1996

In this paper, we present a new approach for the automation of deburring process. An algorithm for teaching skills of a human expert to a robot manipulator is developed. This approach makes use of TSK fuzzy mode that can wxpress a highly nonlinear functional relation with small number of rules. Burr features such as height, width, area, grinding area are extracted from image processing by use of the vision system. Grinding depth, repetitive number and normal grinding force are chosen as control signals representing actions of the human expert. It is verified that our proposed fuzzy model can accurately express the skills of human experts for the deburring process.

• Journal of the Korean Society for Precision Engineering
• /
• v.10 no.4
• /
• pp.200-205
• /
• 1993

In many applications, the typical parallel-jaw end-effector of a robot arm has been remarkably satisfactory. But, it is not adequate for the applications such as complicated manipulation. In the study, a finger with 4 joints (so, having redundancy) was consturcted to investigate the characteristics of an articulated hand. Each joint was driven by one actuator, and the motor torque was transmited to each joint through a tendon-pulley system. In the context, major considerations for hardware design and the method to solve the inverse kinematics of a redundant manipulator were presented. Finally, the basic capabilities of an articulated hand were presented through experiments.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.35 no.4
• /
• pp.347-352
• /
• 2011

The maximum accuracy of position control by using an industrial robot is about $100{\mu}m$, whereas the maximum tolerated imprecision in the position of precision parts is about several tens of micrometers. Therefore, it is very difficult to assemble parts by position control only. Moreover, in the case of precision assembly, jamming or wedging can easily occur because of small position/orientation errors, which may damage the parts to be assembled. To overcome these problems, we investigated a force control scheme that provides proper motion in response to the contact force. In this study, we constructed a force control system that can be easily implemented in a position-controlled manipulator. Impedance control by using an admittance filter was adopted to perform stable contact tasks. It is shown that the precision parts can be assembled well by adopting impedance control and blind search methods.

• Korean Journal of Agricultural Science
• /
• v.23 no.2
• /
• pp.227-232
• /
• 1996

It is a tedious job to handle seedling trays in a green house with human hands because of their fluctuation when they are moving. A robotic gripper for handling a seedling tray has been developed., which has two fingers consisting of the vertical and horizontal bars. The maximum deflections at the center of the seedling trays were measured with various lengths of the horizontal bars. The length of 250 mm was revealed to be optimal one, resulting less than 15 mm deflection with a $540{\times}270mm$ seedling tray, which was acceptable for a practical use. For this study a LM system was adopted to move the fingers. To validate this system the robotic gripper was installed on a Cartesian robotic manipulator and their performance was tested several times with great success. The robotic manipulator used in this study was a general one, so a special one for this robotic gripper needs to be developed in future.

• Journal of Institute of Control, Robotics and Systems
• /
• v.18 no.12
• /
• pp.1106-1114
• /
• 2012

To achieve synchronized motion between a wearable robot and a human user, the redundancy must be resolved in the same manner by both systems. According to the seven DOF (Degrees of Freedom) human arm model composed of the shoulder, elbow, and wrist joints, positioning and orientating the wrist in space is a task requiring only six DOFs. Due to this redundancy, a given task can be completed by multiple arm configurations, and thus there exists no unique mathematical solution to the inverse kinematics. This paper presents analysis on the kinematic and dynamic aspect of the human arm movement and their effect on the redundancy resolution of the human arm based on a seven DOF manipulator model. The redundancy of the arm is expressed mathematically by defining the swivel angle. The final form of swivel angle can be represented as a linear combination of two different swivel angles achieved by optimizing different cost functions based on kinematic and dynamic criteria. The kinematic criterion is to maximize the projection of the longest principal axis of the manipulability ellipsoid for the human arm on the vector connecting the wrist and the virtual target on the head region. The dynamic criterion is to minimize the mechanical work done in the joint space for each two consecutive points along the task space trajectory. As a first step, the redundancy based on the kinematic criterion will be thoroughly studied based on the motion capture data analysis. Experimental results indicate that by using the proposed redundancy resolution criterion in the kinematic level, error between the predicted and the actual swivel angle acquired from the motor control system is less than five degrees.