• International Journal of Naval Architecture and Ocean Engineering
• /
• 제12권1호
• /
• pp.768-783
• /
• 2020

Typically, a Dynamic Positioning System (DPS) uses a PID feed-back system, and it often adopts a wind feed-forward system because of its easier implementation than a feed-forward system based on current or wave. But, because a ship's drifting motion is caused by wind, current, and wave drift loads, all three environmental loads should be considered. In this study, a motion predictive control for the PID feedback system of the DPS is proposed, which considers the three environmental loads by utilizing predicted drifted ship positions in the future since it contains information about the three environmental loads from the moment to the future. The prediction accuracy for the future drifted ship position is ensured by adopting deep learning algorithms and a replay buffer. Finally, it is shown that the proposed motion predictive system results in better station-keeping performance than the wind feed-forward system.

• 한국항해항만학회지
• /
• 제31권7호
• /
• pp.597-603
• /
• 2007

A predictive controller can solve a control problem related to a disturbance-dominant system such as roll stabilization of a ship in waves. In this paper, a predictive controller is developed for a fin stabilizer. Future wave-induced moment is modeled simply using two typical regular wave components for which six parameters are identified by the recursive Fourier transform and the least squares method using the past time series of the roll motion. After predicting the future wave-induced moment, optimal control theory is applied to discover the most effective command fin angle that will stabilize the roll motion. In the results, wave prediction performance is investigated, and the effectiveness of the predictive controller is compared to a conventional PD controller.

• 한국정밀공학회:학술대회논문집
• /
• 한국정밀공학회 1995년도 추계학술대회 논문집
• /
• pp.538-541
• /
• 1995

Joint flexibilities and frictional uncertainties are known to be a major cause of performance degration in motion control systems. This paper investigates the modeling and compensation of these undesired effects. A hybrid controller, which consists of a predictive controller and a neural network controller, is designed to overcome these undesired effects. Also learning scheme for friction uncertainies, which don't interfere with feedback controller dynamics, is discussed. Through simulation works with two inetia-torsional spring system having Coulomb friction, the effectiveness of the proposed hybrid controller was tested. The proposed predictive & neural network hybrid controller shows better performance over one when only predictive controller used.

• 로봇학회논문지
• /
• 제17권3호
• /
• pp.359-364
• /
• 2022

This study presents a nonlinear model predictive control (NMPC)-based obstacle avoidance and whole-body motion planning method for the mobile manipulators. For the whole-body motion control, the mobile manipulator with an omnidirectional mobile base was modeled as a nine degrees-of-freedom (DoFs) serial open chain with the PPR (base) plus 6R (arm) joints, and a swept sphere volume (SSV) was applied to define a convex hull for collision avoidance. The proposed receding horizon control scheme can generate a trajectory to track the end-effector pose while avoiding the self-collision and obstacle in the task space. The proposed method could be calculated using an interior-point (IP) method solver with 100[ms] sampling time and ten samples of horizon size, and the validation of the method was conducted in the environment of Pybullet simulation.

• 한국컴퓨터그래픽스학회논문지
• /
• 제27권1호
• /
• pp.9-17
• /
• 2021

본 논문에서는 확률적 모델예측제어(model predictive control) 기법을 이용하여 예제 동작 데이터가 주어지면 물리 기반 시뮬레이션 환경에서 그 동작을 모방할 수 있는 캐릭터 동작 제어기를 빠르게 학습할 수 있는 간편한 지도 학습(supervised learning) 프레임워크를 제안한다. 제안된 프레임워크는 크게 학습 데이터 생성과 오프라인 학습의 두 컴포넌트로 구성된다. 첫번째 컴포넌트는 예제 동작 데이터가 주어지면 확률적 모델예측제어를 통해 그 동작 데이터를 추적하기 위한 최적 제어기를 캐릭터의 현재 상태로부터 시작하여 가까운 미래 상태까지의 시간 윈도우에 대해 주기적으로 업데이트하면서 그 최적 제어기를 통해 캐릭터의 동작을 확률적으로 제어한다. 이러한 주기적인 최적 제어기의 업데이트와 확률적 제어는 주어진 예제 동작 데이터를 모방하는 동안 캐릭터가 가질 수 있는 다양한 상태들을 효과적으로 탐색하게 하여 지도 학습에 유용한 학습 데이터를 수집할 수 있게 해준다. 이렇게 학습 데이터가 수집되면, 오프라인 학습 컴포넌트에서는 그 수집된 데이터를 정규화 시켜서 데이터에 내제된 크기와 단위의 차이를 조정하고 지도 학습을 통해 제어기를 위한 간단한 구조의 인공 신경망을 학습시킨다. 걷기 동작과 달리기 동작에 대한 실험은 본 논문에서 제안한 학습 프레임워크가 물리 기반 캐릭터 동작 제어기를 빠르고 효과적으로 생성할 수 있음을 보여준다.

• 자동차안전학회지
• /
• 제14권4호
• /
• pp.35-42
• /
• 2022

This paper presents a stochastic model predictive control (SMPC) for stop maneuver of autonomous vehicles considering perception uncertainty of stopped vehicle. The vehicle longitudinal motion should achieve both driving comfortability and safety. The comfortable stop maneuver can be performed by mimicking acceleration profile of human driving pattern. In order to implement human-like stop motion, we propose a reference safe inter-distance and velocity model for the longitudinal control system. The SMPC is used to track the reference model which contains the position uncertainty of preceding vehicle as a chance constraint. We conduct simulation studies of deceleration scenarios against stopped vehicle in urban environment. The test results show that proposed SMPC can execute comfortable stop maneuver and guarantee safety simultaneously.

• 한국컴퓨터그래픽스학회논문지
• /
• 제25권3호
• /
• pp.133-142
• /
• 2019

물리기반 캐릭터 애니메이션에서 궤적 최적화(trajectory optimization) 기법은 캐릭터 동작에 대한 시스템 동역학 모델(system dynamics model)에 기반하여 가까운 최적의 미래 상태를 예측하여 캐릭터의 동작을 자동적으로 생성하는데 널리 사용되어 왔다. 캐릭터와 환경 간의 접촉 현상을 강체 충돌로 다루는 경우 일반적으로 시스템 동역학 모델은 그 수식이 닫힌 형식(closed form)으로 유도되지 못하고 미분이 불가능하다. 따라서 최근까지 많은 연구자들이 접촉 완화(contact smoothing) 기법을 통해 시스템 동역학의 수치적 미분에 기반한 효율적인 궤적 최적화 기법을 발표해 왔다. 하지만 수치적 미분 정보는 분석적 미분과 달리 부정확하기 때문에 궤적 최적화의 안정성에 영향을 미칠 수 있다. 이 문제를 해결하기 위해 본 논문에서는 접촉 완화 모델에 대한 근사화를 통해 시스템 동역학을 분석적으로 미분하여 닫힌 형식의 도함수를 유도하고, 이를 기반으로 사용자의 온라인 입력에 따라 예제 데이터 없이 이족 캐릭터의 동작을 안정적으로 생성하는 예측 제어 기법(model predictive control (MPC))을 제안한다.

• 제어로봇시스템학회논문지
• /
• 제7권12호
• /
• pp.967-972
• /
• 2001

In this paper, an adaptive robust nonlinear predictive controller is developed for the continuous time nonlinear systems whose control objective is composed of the system output and its desired value. The basic control law is derived from the continuous time prediction model and its feedback dynamcis shows another from if input and output linearization. In order to cope with the parameter uncertainty, robust control is incorporated into the basic control law and the asymptotic convergence of tracking error to a certain bounded region is guaranteed. For stability and performance improvement within the bounded region, an adaptive control is introduced. Simulation tests for the motion control of an underwater wall-ranging robot confirm the performance improvement and the robustness of this controller.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1996년도 한국자동제어학술회의논문집(국내학술편); 포항공과대학교, 포항; 24-26 Oct. 1996
• /
• pp.302-305
• /
• 1996

This paper presents a visual servoing combined by Neural Network with optimal structure and predictive control for robotic manipulators to tracking or grasping of the moving object. Using the four feature image information from CCD camera attached to end-effector of RV-M2 robot manipulator having 5 dof, we want to predict the updated position of the object. The Kalman filter is used to estimate the motion parameters, namely the state vector of the moving object in successive image frames, and using the multi layer feedforward neural network that permits the connection of other layers, evolutionary programming(EP) that search the structure and weight of the neural network, and evolution strategies(ES) which training the weight of neuron, we optimized the net structure of control scheme. The validity and effectiveness of the proposed control scheme and predictive control of moving object will be verified by computer simulation.

• 제어로봇시스템학회논문지
• /
• 제20권9호
• /
• pp.957-963
• /
• 2014

This paper proposes a predictive control for an efficient human following robot using Kinect sensor. Especially, this research is focused on detecting of foot-end-point and foot-vector instead of human body which can be occluded easily by the obstacles. Recognition of the foot-end-point by the Kinect sensor is reliable since the two feet images can be utilized, which increases the detection possibility of the human motion. Depth image features and a decision tree have been utilized to estimate the foot end-point precisely. A tracking point average algorithm is also adopted in this research to estimate the location of foot accurately. Using the continuous locations of foot, the human motion trajectory is estimated to guide the mobile robot along a smooth path to the human. It is verified through the experiments that detecting foot-end-point is more reliable and efficient than detecting the human body. Finally, the tracking performance of the mobile robot is demonstrated with a human motion along an 'L' shape course.