• Journal of the Korea Institute of Military Science and Technology
• /
• v.24 no.5
• /
• pp.558-568
• /
• 2021

The existing underwater vehicle controller design is applied by linearizing the nonlinear dynamics model to a specific motion section. Since the linear controller has unstable control performance in a transient state, various studies have been conducted to overcome this problem. Recently, there have been studies to improve the control performance in the transient state by using reinforcement learning. Reinforcement learning can be largely divided into value-based reinforcement learning and policy-based reinforcement learning. In this paper, we propose the roll controller of underwater vehicle based on Deep Deterministic Policy Gradient(DDPG) that learns the control policy and can show stable control performance in various situations and environments. The performance of the proposed DDPG based roll controller was verified through simulation and compared with the existing PID and DQN with Normalized Advantage Functions based roll controllers.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.19 no.7
• /
• pp.1-8
• /
• 2018

This paper describes new communication methods for transmitting torque commands between the vehicle controller that determines the amount of power generation in a range-extended electric vehicle and the engine controller that performs it. Generally, vehicles use CAN communication, but in this case, the hardware and software of the existing engine controller must be modified. For this reason, it is not easy to apply CAN communication to small and medium sized automotive reorganize companies. Therefore, this research presents a pin-pin communication method for applying the existing mass produced engine controller to range-extended electric vehicles. The pin-pin communication method converts the driver's demand torque control map inside an mass produced engine controller into a virtual accelerator opening position according to the target speed and target torque of the engine, and converts this to a voltage signal for the existing mass produced engine controller to recognize it. The virtual accelerator opening positions are mounted in the form of a control map in the vehicle controller through the reverse conversion process in an offline environment and are determined by the engine generating power requirements and engine optimal operating point algorithm. These algorithms and signal conversion circuits for engine torque transmission have been mounted on the vehicle controller to conduct the virtual accelerator opening position conversion process according to the engine target torque and to establish the virtual accelerator voltage signal using the signal converter.

• The Transactions of the Korean Institute of Electrical Engineers D
• /
• v.50 no.8
• /
• pp.361-366
• /
• 2001

This thesis proposes a fuzzy logic cross coupled controller for a multi axis servo system. The overall control system consists of three elements: the axial position controller, the speed controller, and a fuzzy logic cross coupled controller. In conventional multi axis servo system, the motion of each axis is controlled independently without regard to the motion of other axes, in which the contour error, defined as the shortest distance between the desired and actual contours is compensated only by the position error of each axis. This decoupled control approach may result in degraded contouring performance due to such factors as mismatch of axial dynamics and axial loop gains. In practice, such systems contain many uncertainties, Therefore, the multi axis servo system must receive and evaluate the motion of all axes for a better contouring accuracy. Cross coupled controller utilizes all axis position error information simultaneously to produce accurate contours. However the existing cross coupled controllers cannot overcome friction, backlash and parameter variation. Also, since it is difficult to obtain an accurate mathematical model of multi axis system, here we investigate a fuzzy logic cross coupled controller method. Some simulations and experimental results are presented to illustrate the performance of the proposed controller.

• 제어로봇시스템학회:학술대회논문집
• /
• 1991.10a
• /
• pp.1022-1027
• /
• 1991

A new method to control a robot manipulator by neural networks is proposed. The controller is composed of both a PD controller and a neural network-based feedforward controller. MLP(multi-layer perceptron) neural network is used for the feedforward controller and trained by BP(back-propagation) learning rule. Error terms for BP learning rule are composed of the outputs of a PD controller and the acceleration errors of manipulator joints. We compare the proposed method with existing ones and contrast performances of them by simulation. Also, We discuss the real application of the proposed method in consideration of the learning time of the neural network and the time required for sensing the joint acceleration.

• Journal of Institute of Control, Robotics and Systems
• /
• v.20 no.7
• /
• pp.695-700
• /
• 2014

In this paper, we utilize the CMG's momentum bias to control the roll/pitch attitude of the Quad-rotor. While the previous control approaches have used the thrust control approach, we design and add a new momentum controller (using CMG) in order to improve the transient response over the existing methods. The focal point of this paper is the design of a controller for a Quad-rotor's attitude using CMG. This leads to other tasks such as an identification of the model's parameters and mathematical nonlinear modeling. Then, the previous thrust controller is designed based on the linearized model. Finally, the overall system with our designed controller is implemented and tested in real time to show that the Quad-rotor is kept in a good balanced position faster than the traditional thrust-only control approach.

• Journal of Advanced Marine Engineering and Technology
• /
• v.30 no.6
• /
• pp.760-769
• /
• 2006

This paper proposes an autopilot system using a fuzzy PID controller to satisfy performances required for the automatic navigation of ships under various marine circumstances. The existing autopilot system using a PD type controller has difficulties in eliminating a steady-state error and compensating nonlinear characteristics of ships. The autopilot system using the proposed fuzzy PID controller has a self-tuning ability, an ability to compensate nonlinear characteristics, and an ability to turn at constant angular velocity. Therefore. it can naturally make a steady-state error zero, compensate nonlinear dynamic effect of ships, have an adaptability to parameter variation owing to shallow water effect, and have an ability to turn ship's course rapidly without overshoot through procedures of acceleration, constant, and deceleration of angular velocity for large course-changing.

• Proceedings of the KSR Conference
• /
• 2007.11a
• /
• pp.865-871
• /
• 2007

The levitation system of Maglev is composed with electro-magnet, power supplier, controller and sensor. The complex interactions between above subcomponents define the characteristics of electromagnetic suspension of the vehicle. In this study, to understand the influence of controller on the running performance of Maglev, the new controller based on LQ theory will be designed and be simulated with simplified vehicle model. Then the influence of controller on the characteristics of electromagnetic suspension will be reviewed through comparison with existing control algorithm of our prototype vehicle.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.2 no.3
• /
• pp.220-226
• /
• 2002

In this paper, we develop the intelligent digitally redesigned fuzzy controller for nonlinear systems. Takagi-Sugeno (TS) fuzzy model is used to model the nonlinear systems and a continuous-time fuzzy-model-based controller is designed based on the extended parallel-distributed-compensation(EPDC) method . The digital controllers are determined from existing analogue controllers. The proposed method provides an accurate and effective method for digital control of continuous-time nonlinear systems and enables us to efficiently implement a digital controller via the pre-determined continuous-time 75 fuzzy-model-based controller. We have applied the proposed method to the duffing forced oscillation system to show the effectiveness and feasibility of the proposed method.

• 제어로봇시스템학회:학술대회논문집
• /
• 2000.10a
• /
• pp.326-326
• /
• 2000

Continuous time system deadbeat controller(CdbC) has been studied mainly since 1992 especially by Japan researchers. They suggested delay elements. These elements stem from the finite Laplace Transform which is the starting point in deadbeat control system design in continuous time system. Every transfer function is established by these elements. From some conditions such as internal model stability and peasibility of a CdbC controller. unknown polynomials or coefficients can be calculated. In this paper, optimal pole placement of the closed loop system is suggested. From this. a CdbC controller with lower order can be obtained which attains the same level of weighted sensitivity function's H$_{\infty}$ norm used as a measure of the robustness property as existing CdbCs.

• The Transactions of the Korean Institute of Electrical Engineers
• /
• v.39 no.9
• /
• pp.944-955
• /
• 1990

A high-performance robotic controller is proposed for uncertain robots by using an adaptive control method, which guarantees the boundedness of uncertain systems with partially known uncertainty bounds. In order to improve the tracking performance of the robotic controller, a linear compensator is introduced to the robotic system which has been linearized via a nonlinear feedback. The above adaptive method is then utilized to guarantee the ultimate boundedness of the tracking errors. The performance of the robotic controller is compared with that of the computed torque method by computer simulations under uncertain environments. The simulation results show that the proposed method gives better performance than the computed torque method. Since the proposed method has a small number of parameters to be estimated, the controller is simpler to implement than other existing adaptive controller for robots. Hence, the proposed robotic control method is expected to be well suited for high-performance operation of robots under uncertain environment.