• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1989년도 한국자동제어학술회의논문집; Seoul, Korea; 27-28 Oct. 1989
• /
• pp.1113-1119
• /
• 1989

Neural network control has many innovative potentials for intelligent adaptive control. Among many, it promises real time adaption, robustness, fault tolerance, and self-learning which can be achieved with little or no system models. In this paper, a dynamic robot controller has been developed based on a backpropagation neural network. It gradually learns the robot's dynamic properties through repetitive movements being initially trained with a PD controller. Its control performance has been tested on a simulated PUMA 560 demonstrating fast learning and convergence.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1991년도 한국자동제어학술회의논문집(국내학술편); KOEX, Seoul; 22-24 Oct. 1991
• /
• pp.557-562
• /
• 1991

In this paper, we present new methods for solving the inverse kinematics associated with 6 degree of freedoms manipulator by the numerical method. This method will be based on tracking stability of special nonlinear dynamical systems, and differs from the typical techniques based by the Newton-Gauss or Newton-Raphson method for solving nonlinear equations. This simulation results show that the new method is solving the inverse kinematics of PUMA 560 without the derivative of a given task space trajectories.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1989년도 한국자동제어학술회의논문집; Seoul, Korea; 27-28 Oct. 1989
• /
• pp.244-249
• /
• 1989

For the robot manipulator in performing precision tasks, it is indispensable that the robot utilize the various sensors for intelligence. In this paper, the hybrid position/force control method is implemented with a force/torque sensor, two personal computers, and a PUMA 560 manipulator. Two application S/W packages for edge following and peg-in-hole tasks are developed by the proposed force control algorithm. The related experimental results are then presented and discussed,

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1988년도 한국자동제어학술회의논문집(국내학술편); 한국전력공사연수원, 서울; 21-22 Oct. 1988
• /
• pp.67-71
• /
• 1988

We present an algorithm that uses trajectory following errors to improve a feedforward command to a robot in the iterative manner. It has been shown that when the manipulator handles an unknown object, the P-type learning algorithm can make the trajectory converge to a desired path and also that the proposed learning control algorithm performs better than the other type learning control algorithm. A numerical simulation of a three degree of freedom manipulator such as PUMA-560 ROBOT has been performed to illustrate the effectiveness of the proposed learning algorithm.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1990년도 한국자동제어학술회의논문집(국내학술편); KOEX, Seoul; 26-27 Oct. 1990
• /
• pp.24-29
• /
• 1990

In this paper ,ie propose a hybrid position/force controller of a robot manipulator using double-layer neural network. Each layer is constructed from inverse dynamics and Jacobian transpose matrix, respectively. The weighting value of each neuron is trained by using a feedback force as an error signal. If the neural networks are sufficiently trained it does not require the feedback-loop with error signals. The effectiveness of the proposed hybrid position/force controller is demonstrated by computer simulation using a PUMA 560 manipulator.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1990년도 한국자동제어학술회의논문집(국내학술편); KOEX, Seoul; 26-27 Oct. 1990
• /
• pp.608-612
• /
• 1990

The robot system which can handle a stream of randomly positioned parts on a conveyor belt system, is developed. It is composed of a PUMA 560 robot, a conveyor belt system and a vision system. The performance of the overall system is mainly dependent upon the robot and vision system interface technique. A vision algorithm is developed to determine the position, orientation and type of the part. Calibration procedure and the vision-to-robot transformation are also proposed. Experimental results are then presented and discussed.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 1988년도 전기.전자공학 학술대회 논문집
• /
• pp.998-1001
• /
• 1988

A robust control method for robots in presented. In this method, a phase-lag controller is used for reducing the effect of the unknown payload without the measurement of joint accelerations and torque/force. Simulation results for the lower 3 joints of PUMA 560 show considerable reduction of position errors due to the unknown payload, compared to the computed-torque method.

• 전자공학회논문지B
• /
• 제28B권11호
• /
• pp.897-903
• /
• 1991

In this paper we propose a hybrid position/force controller of a robot manipulator using feedback error learning rule and neural networks. The neural network is constructed from inverse dynamics. The weighting value of each neuron is trained by using a feedback force as an error signal. If the neural networks are sufficiently trained well, it does not require the feedback-loop with error signals. The effectiveness of the proposed hybrid position/force controller is demonstrated by computer simulation using PUMA 560 manipulator.

• 전자공학회논문지SC
• /
• 제43권5호
• /
• pp.10-18
• /
• 2006

본 논문에서는 다중 로봇 시스템에서 물체와 로봇 팔끝 간에 접촉 마찰이 존재할 때 이 로봇 시스템의 조작도를 해석하는 새로운 방법을 제안한다. 로봇이 물체를 떨어뜨리지 않고 잡고 있으려면, 로봇이 물체에 가하는 힘 벡터가 friction cone 내부에 존재 해야만 한다. 이러한 friction cone 내부를 나타내는 식은 일반적으로 비선형 형태로 되어 있기 때문에 기존의 조작도 분석 방법에 이 식을 구속 조건으로 적용하기가 쉽지 않다. 따라서 본 논문에서는 이러한 friction cone 내부를 다각뿔로 근사함으로써 선형적인 구속 조건으로 표현하였다. 또한 선행 연구에서 찾지 못했던 부분을 새롭게 찾아내었다. 그리고 다중 로봇 시스템에 조작도를 나타내는 물체 중심의 가속도를 구하기 위해서, 먼저 선형계획법을 통해서 허용 가능한 토크의 영역을 구하였다. 이 토크의 영역을 선형 변환을 통해 최종적으로 물체의 최대 가속도의 영역을 구하였다. 본 방법의 타당성을 입증하기 위해서 두 대로 구성 된 다중 로봇 시스템과 PUMA560 로봇 시스템에 적용하였다.

• 전자공학회논문지B
• /
• 제29B권8호
• /
• pp.15-23
• /
• 1992

Neural network control has many innovative potentials for fast, accurate and intelligent adaptive control. In this paper, two kinds of neurocontrol architectures for the dynamic control of robot manipulators are developed. One is based on a System Identification and Control scheme and the other is based on the Feedback-Error leaming scheme. Both of the proposed architectures use an inverse dynamic neurocontroller in parallel with a linear neurocontroller. The difference is that the first architecture uses the system identifier to get the signals used for training neurocontrollers, while the second architecture uses a properly defined energy function. Compared with the previous types of neurocontrollers which are using an inverse dynamic neurocontroller and a fixed PD gain controller, the proposed architectures not only eliminate the painful process of the fixed gain tuning but also exhibit superior peformances because the linear neurocontroller can adapt its gains according to the applied task. This superior performance is tested and verified through computer simulation of the dynamic control of the PUMA 560 arm.