• Journal of the Korea Computer Graphics Society
• /
• v.30 no.3
• /
• pp.109-123
• /
• 2024

Research into vision-based end-to-end autonomous driving systems utilizing deep learning and reinforcement learning has been steadily increasing. These systems typically encode continuous and high-dimensional vehicle states, such as location, velocity, orientation, and sensor data, into latent features, which are then decoded into a vehicular control policy. The complexity of urban driving environments necessitates the use of state representation learning through networks like Variational Autoencoders (VAEs) or Convolutional Neural Networks (CNNs). This paper analyzes the impact of different image state encoding methods on reinforcement learning performance in autonomous driving. Experiments were conducted in the CARLA simulator using RGB images and semantically segmented images captured by the vehicle's front camera. These images were encoded using VAE and Vision Transformer (ViT) networks. The study examines how these networks influence the agents' learning outcomes and experimentally demonstrates the role of each state representation technique in enhancing the learning efficiency and decision- making capabilities of autonomous driving systems.

• 제어로봇시스템학회:학술대회논문집
• /
• 1997.10a
• /
• pp.272-275
• /
• 1997

When a manipulator makes contact with an object having position uncertainty, performance measures vary considerably with the control law. To achieve the optimal solution for this problem, an unique objective function that weights time and impact force is suggested and is solved with the help of variational calculus. The resulting optimal velocity profile is then modified to define a sliding mode for the impact and force control. The sliding mode control technique is used to achieve the desired performance. Sets of experiments are performed, which show superior performance compared to any existing controller.

• 제어로봇시스템학회:학술대회논문집
• /
• 1993.10b
• /
• pp.155-159
• /
• 1993

In this paper, we consider the problem of a stochastic optimal switching control, which can be applied to the control of a system with uncertain demand such as a control problem of a power plant. The dynamic programming method is applied for the formulation of the optimal control problem. We solve the system of Quasi-Variational Inequalities(QVI) using an algoritlim which involves the finite difference approximation and contraction mapping method. A mathematical example of the optimal switching control is constructed. The actual performance of the algorithm is also tested through the solution of the constructed example.

• Journal of applied mathematics & informatics
• /
• v.6 no.2
• /
• pp.631-638
• /
• 1999

Sensitive testing has been widely employed for many years in connection with the development and evaluation explosives detonation devices and propellants. Perhaps its earliest and possibly most important implementation was in biological studies of dosage mortality and response to drugs. Recently sensitivity experiments has been employed in the evaluation of new materials subject to stress in various environments and in delineanation of unstable combustion regions in chemical propulsion systems. This paper discussed a sta-tistical development of sensitivity testing.

• 제어로봇시스템학회:학술대회논문집
• /
• 1988.10a
• /
• pp.50-53
• /
• 1988

This paper presents an approach to the position control of a robot manipulator by using a self-tuning pole-placement controller with an inverse model. The linearized independent difference equations of manipulator motion are obtained, and the parameters of these models are estimated on line. The controller is composed of two parts, the primary controller obtains desired torques by using an inverse model and the secondary controller computes variational torques on the basis of induced perturbation equations by minimizing a quadratic criterion with a closed-loop pole-placement. Simulation is performed to demonstrate the effectiveness of this approach.

• 제어로봇시스템학회:학술대회논문집
• /
• 1986.10a
• /
• pp.506-509
• /
• 1986

In this paper, the control law for the system that has the variation of moment of inertia is designed. The proposed method is that the control input is obtained by using optimal PI control and partial adaptive control. The partial adaptive control input is adjusted by estimating the variational quantity of moment of inertia. This result gives us significant improvement of tracking ability.

• Progress in Superconductivity
• /
• v.11 no.2
• /
• pp.135-140
• /
• 2010

We propose a variational wave function for the ground state of the magnetic heavy fermion (HF) systems, in which both the Kondo and the RKKY interactions are variationally incorporated and the local f-orbital state exists as a linear combination of a full local moment state and a fully compensated state (mixed wave state). We describe the mechanism for the mixed wave ground state based on the large-N treatment of the Kondo lattice Hamiltonian added with RKKY interaction. With the mixed wave ground state we can explain several puzzling experiments in magnetic HF compounds such as a small value of local moment, coexistence of the antiferromagnetic (AFM) and the paramagnetic (PM) phases, local quantum criticality, etc.

• Journal of applied mathematics & informatics
• /
• v.36 no.5_6
• /
• pp.491-504
• /
• 2018

In this article, we consider a fourth order nonlinear difference system. By making use of the critical point theory, we obtain some new existence theorems of at least one periodic solution with minimal period. Our main approach used in this article is the variational technique and the Saddle Point Theorem.

• Journal of applied mathematics & informatics
• /
• v.23 no.1_2
• /
• pp.547-552
• /
• 2007

In this note, we extends some of the results of Liu [Fuzzy Sets and systems 157 (2006) 869-878]. This extension consists of a simple proof involving weighted functions and their preference index. We also give an elementary simple proof of the maximum entropy weighting function problem with a given preference index value without using any advanced theory like variational principles or without using Lagrangian multiplier methods.

• The Pure and Applied Mathematics
• /
• v.3 no.1
• /
• pp.83-94
• /
• 1996

Classical mechanics begins with some variants of Newton's laws. Lagrangian mechanics describes motion of a mechanical system in the configuration space which is a differential manifold defined by holonomic constraints. For a conservative system, the equations of motion are derived from the Lagrangian function on Hamilton's variational principle as a system of the second order differential equations. Thus, for conservative systems, Newtonian mechanics is a particular case of Lagrangian mechanics.(omitted)