|
Learning-based Predictive Control for Vehicle Following Problems |
|
Ziliang Wang, Shuyou Yu, Yajing Zhang, Xiaohui Yu(Jilin University, China), Hong Chen(Tongji University, China) |
| Recent research shows that the combination of reinforcement learning (RL) with traditional control method can be an effective tool for designing near optimal feedback controller for dynamic systems.In this paper, a vehicle-following control based on reinforcement learning is proposed, in which pairs of the input-output of model predictive control (MPC) are chosen as offline-learning data.Through continuous iterations of actor-network and critic-network,the longitudinal vehicle-following controller can be obtained. Simulation results illustrate that proposed learning-based predictive control (LPC) can improve the computational efficiency,and obtain a better performance of policy optimization. |
|
|
Improving Time Delay Control Performance for Varying Disturbance using Gaussian Process Extrapolation |
|
Jung-Su Kim(SeoulTech, Korea), Abdul Aris Umar(Seoul National University of Science and Technology, Korea) |
| The stability of a system highly depends on the robustness of the applied control method. One of the straightforward
yet powerful methods to ensure closed-loop system robustness is time delay control. However, the estimated
disturbance of this method suffers from a lagging phenomenon when the unknown external force is not constant or the
sampling time of the system is not fast enough. To this end, this paper proposes an extrapolation method using Gaussian
process regression to improve the disturbance estimation of time delay control. A spectral mixture kernel is employed as
it plays a vital role in Gaussian process extrapolation. Simulation results are presented to show the effectiveness of |
|
|
Characteristics of Autopilot Lag and Time-Varying Thrust for Model Predictive Control based Impact Angle Guidance Law |
|
Sungjin Cho(Sunchon National University, Korea) |
| This paper presents characteristics of autopilot lag and time-varying thrust for a model predictive control (MPC)
based guidance law. Since UAV speed continues to changing because of time-varying thrust, recent research work develops analytical solution based impact angle guidance law for time-varying thrust. However, guidance performance can be degraded when autopilot lag is included in the guidance law. Hence, we analyze characteristics of autopilot lag and time-varying thrust by designing a MPC based impact angle guidance law. |
|
|
Current constrained single-loop model predictive voltage control for DC-DC buck converter using online optimization |
|
Ngoc-Duc Nguyen, Young Il Lee(SeoulTech, Korea) |
| - A single-loop model predicitve voltage controller (MPVC) for the synchronous buck DC-DC converter is proposed using MPC technique considering the current constraint.
- The proposed MPVC can have a non-overshoot regulation in both the inductor current and output voltage without violating the current limit.
- Comparing the MPVC with the typical PI controller, the MPVC can have a better voltage regulation not only at the start-up but also under the full-load condition. |
|