The CDSL workshop was held in Jeju International Convention Center in from October 17 to 18 with the participation of various laboratories for the study of control theory. In particular, Prof. Juhoon Back (Kwangwoon Univ.), Prof. Nam H. Jo (Soongsil Univ.), and the members of Distributed Control and Autonomous Systems Lab. graced the workshop with their presence and participation.
The workshop took place over two days. On the first day, it was conducted with five practice talks to be presented at 58th IEEE Conference on Decision and Control, and on the second day, conducted with three oral presentations and six poster presentations that the students of CDSL have studied. The details of each talk can be found in the table below.
There were many good presentations, but here let us briefly introduce one interesting presentation done by Seongjun Kim. He is not only attending Ph.D. program in CDSL, but also working at Agency for Defense Development (ADD). As a projectile researcher, he presents the general modelling techniques of systems, such as missiles, and basic guided steering concepts for matching targets, and how to deal with some problematic situations that may arise in real-world environments.
Prof. Nam H. Jo said, “Thank you for giving us the opportunity to learn the core of new research topics being conducted in CDSL in a short time, and I’m already waiting for the event next year.”
Control and Dynamic Systems Laboratory (CDSL) has held the workshop every year. Professor Shim, head of CDSL, said, “Our workshop is planned to encourage active discussion on control theory among researchers. Hopefully, it constitutes a strong network between researchers in order to enhance their research and knowledge”.
List of presentations
Oct 17 (Friday)
- Encrypted State Estimation in Networked Control Systems, Junsoo Kim
Abstract : In this paper, we propose a distributed state observer in which the data of measurements and state estimates are protected by homomorphic cryptosystem. The proposed observer network is composed of local observers, where each of them utilizes encrypted local measurement, encryption of plant input, and encrypted estimates transmitted from its neighbors. All the operations in the network are performed on encrypted data without decryption, and the full state is recovered in every local observer as an encrypted message. Assuming the characteristic polynomial and the minimal polynomial of the state matrix for the plant are the same, the parameters in the observers are chosen to be integers. This not only allows finite time convergence for the state estimates, but also makes the encrypted dynamic observers operate for infinite time horizon. - On Improving the Robustness of Reinforcement Learning-based Controllers using Disturbance Observer, Jeong Woo Kim
Abstract : Because reinforcement learning (RL) may cause issues in stability and safety when directly applied to physical systems, a simulator is often used to learn a control policy. However, the control performance may be easily deteriorated in a real plant due to the discrepancy between the simulator and the plant. In this paper, we propose an idea to enhance the robustness of such RL-based controllers by utilizing the disturbance observer (DOB). This method compensates for the mismatch between the plant and simulator, and rejects disturbance to maintain the nominal performance while guaranteeing robust stability. Furthermore, the proposed approach can be applied to partially observable systems. We also characterize conditions under which the learned controller has a provable performance bound when connected to the physical system. - Distributed Algorithm for Economic Dispatch Problem with Separable Losses, Seungjoon Lee
Abstract : Economic dispatch problem for a networked power system has been considered. The objective is to minimize the total generation cost while meeting the overall supply-demand balance and generation capacity. In particular, a more practical scenario has been studied by considering the power losses. A non-convex optimization problem has been formulated where the non-convexity comes from the nonlinear equality constraint representing the supply-demand balance with the power losses. It is shown that the optimization problem can be solved using convex relaxation and dual decomposition. A simple distributed algorithm is proposed to solve the optimization problem. Specifically, the proposed algorithm does not require any initialization process and hence robust to various changes in operating condition. In addition, the behavior of the proposed algorithm is analyzed when the problem is infeasible. - Study on Realizable Generalized Hold Functions as a Countermeasure against Zero Dynamics Attack, Jongsoo Ha
Abstract : Zero dynamics attacks are known to be lethal in the sense that they are stealthy in principle and are not detected from output measurements. Therefore, instead of detecting the zero dynamics attacks, an idea to mitigate the effect of the zero dynamics attack has been proposed recently, which is to enforce the zeros to become stable by changing the zero-order hold to a generalized hold in the sampled data framework. Once all the zeros become stable, then even if the zero dynamics attack is engaged, its effect on the plant is negligible. However, it was observed that the amplitude of the generalized hold becomes unrealistically large in some cases, which leads to a large input to the physical plant. This paper studies this phenomenon at a deeper level and figures out that changing the intrinsic zeros requires an excessively large amplitude of the generalized hold while changing the sampling zeros can be done with a reasonable amplitude. - Distributed Kalman-filtering: Distributed optimization viewpoint, Kunhee Ryu (Kwangwoon Univ)
Abstract : In this paper, we propose a novel computation model for solving the distributed optimization problem where the objective function is formed by the sum of convex functions available to individual agent. Our approach differentiates from the existing approach by local convex mixing and gradient searching in that we force the states of the model to the global optimal point by controlling the subgradient of the global optimal function. In this way, the model we proposed does not suffer from the limitation of diminishing step size in gradient searching and allows fast asymptotic convergence. The model also shows robustness to additive noise, which is a main curse for algorithms based on convex mixing or consensus.
Oct 18 (Friday)
- DO-DAT: A MATLAB toolbox for design and analysis of disturbance observer, Hamin Chang
Abstract : In this presentation, we provide a new necessary and sufficient condition for the robust stability of the DOB controlled system. In particular, our study proposes a new method to design the conventional DOB by providing a proper range of bandwidth of the Q-filter included in the DOB. A new version of DO-DAT (a MATLAB toolbox) that reflects all the contents of this presentation will be available soon. - Weighted Averaged Behavior of Synchronization between Heterogeneous Agents in a Sampled-data Setting, Jiyeon Nam
- An Introduction to the guided missile, Seongjun Kim
- Simulation Result on Design of Discrete PFC on Sampled Data System under System Uncertainty, Joowon Lee
Abstract : It is almost impossible to obtain transfer function coefficients of zero-order hold (ZOH) equivalent models exactly without knowing the eigenvalues of system matrices, since it is required to calculate matrix exponentials of system matrices. However, there is no general method of obtaining roots for higher order polynomials. Therefore, we approximate Taylor expansions of ZOH equivalent models to compute transfer function coefficients, and analyze how accurate the approximation is with regard to the sampling period. - Design of Disturbance Observer for Non-minimum Phase Plant with Vibrational Controller, Soojeong Hyun
Abstract : In this poster, we propose a robust feed-through compensator using disturbance observer (DOB). Feed-through compensator induces a new input-output pair so that it has stable zero dynamics and relative degree 1. We define an auxiliary system from the original system. The compensator mainly designed from an output feedback stabilizer for the auxiliary system. We consider a special class of systems with uncertainty. We construct a robust output feedback stabilizer using DOB. Then we finally suggest a compensator and show re-defined output has some desired properties. - Cooperative Observation and Control Recovering Linear Quadratic Gaussian Performance, Donggil Lee
Abstract : This paper introduces a framework of cooperative LQG control based on the networked observers. The main idea of this control approach is: 1) A network of Kalamn filter is designed to cooperatively estimate the state of the plant. 2) The combination of the observers’ feedback based on LQR is used to stabilize the plant. - Shifting intrinsic zeros-TTS form analysis, Jihan Kim
Abstract : Noting the fact that the effectiveness of the zero-dynamics attack heavily relies on the non-minimum phasesness of the system, we had proposed to replace the zero-order hold by the generalized hold to move the unstable zeros into stable ones. However, in order to shift intrinsic zeros, it requires a large hold gain of the generalized hold function. Hence, in this work, by using the truncated Taylor series approximation method, we examine the minimal cost for shifting intrinsic zeros. Moreover, when the relative degree is one, we provide a possible zero assignment region under the minimal cost hold, which is represented by the form of the output feedback closed-loop system. - Vibrational Controller with Uncertain Zero Dynamics, Hyuntae Kim
Abstract : This poster presents a new structure of a sliding mode based controller for handling model uncertainty and unmodeled disturbance. The sliding mode based controller can achieve robust disturbance rejection and nominal performance recovery in finite time. The stability and the finite time convergence of the proposed sliding mode disturbance observer (SM-DOB) are rigorously proved by Lyapunov stability method. - Consensus of Passive Agents, Jeong Mo Seong
Abstract : Preceding researches showed that an output synchronization of passive multi-agent systems is achieved with a static output diffusive coupling protocol for a strongly connected digraph. However, two questions still remain: “what is the synchronized output?” and “Can it be extended to a digraph that has a spanning tree?”. Based on these two questions, the output synchronization of passive multi-agent systems is considered.
Comments are closed.