• 한국기계기술학회지
• /
• 제13권3호
• /
• pp.7-16
• /
• 2011

This study proposes the compensation method for the mechanical deflection error of a SCARA robot. While most studies on the related subject have dealt with the development of a control algorithm for improvement of robot accuracy, this study presents the control method reflecting the mechanical deflection error which is predicted in advance. The deflection at the end of the gripper of SCARA robot is caused by the self-weights and payloads of Arm 1, Arm 2 and quill. If the deflection is constant even though robot's posture and payload vary, there may not be a big problem on robot accuracy because repetitive accuracy, that is relative accuracy, is more important than absolute accuracy in robot. The deflection in the end of the gripper varies as robot's posture and payload change. That's why the moments $M_x$, $M_y$ and $M_z$ working on every joint of a robot vary with robot's posture and payload size. This study suggests the compensation method which predicts the deflection in advance with the variations in robot's posture and payload using neural network. To do this, I chose the posture of robot and the payloads at random, found the deflections by the FEM analysis, and then on the basis of this data, made compensation possible by predicting deflections in advance successively with the variations in robot's posture and payload through neural network learning.

• 제어로봇시스템학회논문지
• /
• 제5권7호
• /
• pp.841-847
• /
• 1999

For successful assembly of flexible parts, informations about their deformation as well as possible misalignments between the holes and their mating parts are essential. Such informations can be acquired from visual sensors. For robotic assembly, the corrective assembly motion to compensate for such misalignments has to be determined from the measured informations. However, this may not be simply derived from the measured misalignment alone because the part deformation progressively occurs during misalignment compensation. Based on the analysis of flexible parts assembly process, this paper presents a neural net-based inference system that can infer the complex relationship between the corrective motion and the measured information of parts deformation and misalignments. And it verifies the performance of the implemented inference system. The results show that the proposed neural net-based misalignment compensation algorithm Is effective in compensating for the lateral misalignment, and that it can be extended to the assembly tasks under more general conditions.

• 대한전기학회논문지:시스템및제어부문D
• /
• 제51권6호
• /
• pp.221-227
• /
• 2002

In this paper, the system model for the compensation of the low-cost personal navigation system is derived and the error compensation method using GPS is also proposed. WNS(Walking Navigation System) is a kind of personal navigation system using the number of a walk, stride and azimuth. Because the accuracy of these variables determines the navigation performance, computational methods have been investigated. The step is detected using the walking patterns, stride is determined by neural network and azimuth is calculated with gyro output. The neural network filters off unnecessary motions. However, the error compensation method is needed, because the error of navigation information increases with time. In this paper, the accumulated error due to the step detection error, stride error and gyro bias is compensated by the integrating with GPS. Loosely coupled Kalman filter is used for the integration of WNS and GPS. It is shown by simulation that the error is bounded even though GPS signal is blocked.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1996년도 한국자동제어학술회의논문집(국내학술편); 포항공과대학교, 포항; 24-26 Oct. 1996
• /
• pp.698-701
• /
• 1996

A novel neural network control scheme is proposed to identify the inverse dynamic model of robot manipulator and to compensate for uncertainties in robot dynamics. The proposed controller is called reference compensation technique(RCT) by compensating at reference input trajectory. The proposed RCT scheme has many benefits due to the differences in compensating position and learning algorithm. Since the compensation is done outside the plant it can be applied to many control systems without modifying the inside controller. It performs well with low controller gain because the operating range of input values is small and the output of the neural network controller is amplified through the controller gain. The back-propagation algorithm is used to train and simulations of three link robot manipulator are carried out to prove the proposed controller's performances.

• 제어로봇시스템학회논문지
• /
• 제10권4호
• /
• pp.341-349
• /
• 2004

In this paper, the decentralized neural network control of the reference compensation technique is proposed to control a 2-DOF inverted pendulum on an x-y plane. The cart with the 2-DOF inverted pendulum moves on the x-y plane and the 2-DOF inverted pendulum rotates freely on the x-y axis. Since the 2-DOF inverted pendulum is divided into two 1-DOF inverted pendulums, the decentralized neural network control is applied not only to balance the angle of pendulum, but also to control the position tracking of the cart. Especially, a circular trajectory tracking is tested for position tracking control of the cart while maintaining the angle of the pendulum. Experimental results show that position control of the inverted pendulum system is successful.

• 제어로봇시스템학회논문지
• /
• 제15권7호
• /
• pp.728-733
• /
• 2009

This paper presents the compensation of mechanical deflection error in SCARA robot. End of robot gripper is deflected by weight of arm and pay-load. If end of robot gripper is deflected constantly regardless of robot configuration, it is not necessary to consider above mechanical deflection error. However, deflection in end of gripper varies because that moment of each axis varies when robot moves, it affects the relative accuracy. I propose the compensation method of deflection error using neural network. FEM analysis to obtain the deflection of gripper end was carried out on various joint angle, the results is used in neural network teaming. The result by simulation showed that maximum relative accuracy reduced maximum 9.48% on a given working area.

• 전력전자학회:학술대회논문집
• /
• 전력전자학회 2001년도 전력전자학술대회 논문집
• /
• pp.389-392
• /
• 2001

This paper presents neural load disturbance observer that used to deadbeat load torque observer and regulation of the compensation gain by parameter estimator. As a result, the response of PMSM follows that of the nominal plant. The load torque compensation method is compose of a neural deadbeat observer. To reduce of the noise effect, the post-filter, which is implemented by MA process, is proposed. The parameter compensator with RLSM (recursive least square method) parameter estimator is suggested to increase the performance of the load torque observer and main controller. The proposed estimator is combined with a high performance neural torque observer to resolve the problems. As a result, the proposed control system becomes a robust and precise system against the load torque and the parameter variation. A stability and usefulness, through the verified computer simulation, are shown in this paper.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• 제8권1호
• /
• pp.68-74
• /
• 2008

In this paper, a novel neural fuzzy control method is proposed to control nonlinear systems. A standard PD-like fuzzy controller is designed and used as a main controller for the system. Then a neural network controller is added to the reference trajectories to form a neural-fuzzy control structure and used to compensate for nonlinear effects. Two neural-fuzzy control schemes based on two well-known neural network control schemes, the feedback error learning scheme and the reference compensation technique scheme as well as the standard PD-like fuzzy control are studied. Those schemes are tested to control the angle and the position of the inverted pendulum and their performances are compared.

• International Journal of Control, Automation, and Systems
• /
• 제4권5호
• /
• pp.545-558
• /
• 2006

In this paper, PD-like visual servoing is modified in two ways: a sliding-mode observer is applied to estimate the joint velocities, and a RBF neural network is used to compensate the unknown gravity and friction. Based on Lyapunov method and input--to-state stability theory, we prove that PD-like visual servoing with the sliding mode observer and the neuro compensator is robust stable when the gain of the PD controller is bigger than the upper bounds of the uncertainties. Several simulations are presented to support the theory results.

• 한국음향학회지
• /
• 제37권6호
• /
• pp.506-511
• /
• 2018

본 논문에서는 잡음 환경에서 효과적인 음성인식을 위하여 GMM(Gaussian Mixture Model)-DNN(Deep Neural Network) 하이브리드 기반의 특징 보상 기법을 제안한다. 기존의 GMM 기반의 특징 보상에서 필요로 하는 사후 확률을 DNN을 통해 계산한다. Aurora 2.0 데이터를 이용한 음성 인식 성능 평가에서 본 논문에서 제안한 GMM-DNN 하이브리드 기법이 기존의 GMM 기반 기법에 비해 Known, Unknown 잡음 환경에서 모두 평균적으로 우수한 성능을 나타낸다. 특히 Unknown 잡음 환경에서 평균 오류율이 9.13 %의 상대 향상률을 나타내고, 낮은 SNR(Signal to Noise Ratio) 잡음 환경에서 상당히 우수한 성능을 보인다.