• Journal of Advanced Research in Ocean Engineering
• /
• 제3권3호
• /
• pp.136-148
• /
• 2017

The paper presents dual arm ROV manipulation using deep reinforcement learning. The purpose of this underwater manipulator is to investigate and excavate natural resources in ocean, finding lost aircraft blackboxes and for performing other extremely dangerous tasks without endangering humans. This research work emphasizes on a self-learning approach using Deep Reinforcement Learning (DRL). DRL technique allows ROV to learn the policy of performing manipulation task directly, from raw image data. Our proposed architecture maps the visual inputs (images) to control actions (output) and get reward after each action, which allows an agent to learn manipulation skill through trial and error method. We have trained our network in simulation. The raw images and rewards are directly provided by our simple Lua simulator. Our simulator achieve accuracy by considering underwater dynamic environmental conditions. Major goal of this research is to provide a smart self-learning way to achieve manipulation in highly dynamic underwater environment. The results showed that a dual robotic arm trained for a 3DOF movement successfully achieved target reaching task in a 2D space by considering real environmental factor.

• 한국지능시스템학회논문지
• /
• 제18권5호
• /
• pp.642-648
• /
• 2008

여자유도 매니퓰레이터의 동적 제어는 관절에 가해지는 토크를 최소화하는 목적으로 많은 연구가 이루어져 왔다. 그러나 기존의 국소 토크 최적화의 동적 제어 방법은 드라이버로 구현하기 힘든 토크가 요구된다. 본 논문에서는 그러한 큰 토크 요구를 상당히 개선시킨 새로운 제어 알고리즘을 제안한다. 이 알고리즘은 기존의 국소 토크 최소화 알고리즘에 퍼지 로직과 유전자 알고리즘을 적용시킨 것이다. 제안된 알고리즘은 3자유도 평면 여자유도 로봇에 적용하였으며, 시뮬레이션 결과를 통하여 제안된 알고리즘의 타당성을 확인하였다.

• 센서학회지
• /
• 제8권1호
• /
• pp.80-88
• /
• 1999

본 논문에서는 로봇의 3차원 작업을 위한 센서로 사용된 스테레오 카메라의 위치 결정 문제를 다룬다. 공통의 기준선상에 평행한 시선을 가지도록 설치된 스테레오 카메라의 모델이 주어진 후, 보정에 사용된 제어점들의 불확실성에 둔감하고 로봇의 반복정밀도를 고려한 오차 조건을 만족시킬 수 있도록 센서의 계측거리가 결정된다. 두 카메라간의 간격은 3차원 위치 오차와 스테레오 영상좌표 오차와의 관계를 고려하여 이들이 최소화될 수 있도록 결정하였다. 본 논문에서 제안한 방법은 기존의 기법들과는 달리 3차원의 문제를 피계측체의 모델링 과정이나 복잡한 제한조건 없이 접근함으로써 일반적이며, 모의실험을 통하여 유용함을 확인할 수 있었다.

• Journal of Welding and Joining
• /
• 제14권6호
• /
• pp.131-139
• /
• 1996

The response of the arc sensor using the welding current and/or welding voltage as its outputs has been obtained by the analysis and/or experiments of the static characteristics of arc sensor. But in order to improve the reliability of arc sensor, it is necessary to know its dynamic characteristics. So in this paper, it is presented the dynamic model of arc sensor including the power source, arc voltage, electrode burnoff rate, and wire feed rate. A numerical simulation of the dynamic model of arc sensor was implemented, computing the welding current with input of CTWD. The results of computer simulations and experiments of $CO_2$arc welding showed that a linear relationship between weaving center - weld line distance and current area difference was established. Additionally, a real-time weld seam tracking system interfaced with industrial welding robot was constructed, the result of the weld seam tracking experiment for weld line with an initial offset error of 5$^{\circ}$was good.

• 제어로봇시스템학회논문지
• /
• 제2권4호
• /
• pp.279-286
• /
• 1996

In this paper, variable structure control(VSC) based on reaching law method with fuzzy inference for nonlinear systems is proposed. The reaching law means the reaching condition which forces an initial state of system to reach switching surface in finite time, and specifies the dynamics of a desired switching function. Since the conventional reaching law has fixed coefficients, the chattering can be existed largely in sliding mode. In the design of a proposed fuzzy reaching law, we fuzzify RP(representative point)'s orthogonal distance to switching surface and RP's distance the origin of the 2-dimensional space whose coordinates are the error and the error rate. The coefficients of the reaching law are varied appropriately by the fuzzy inference. Hence the state of system in reaching mode reaches fastly switching surface by the large values of reaching coefficients and the chattering is reduced in sliding mode by the small values of those. And the effectiveness of the proposed fuzzy reaching law method is showen by the simulation results of the control of a two link robot manipulator.

• Journal of Mechanical Science and Technology
• /
• 제18권4호
• /
• pp.630-639
• /
• 2004

The Stewart platform manipulator is a closed-kinematics chain robot manipulator that is capable of providing high structural rigidity and positional accuracy. However, this is a complex and nonlinear system, so the control performance of the system is not so good. In this paper, a new robust motion control algorithm is proposed. The algorithm uses partial state feedback for a class of nonlinear systems with modeling uncertainties and external disturbances. The major contribution is the design of a robust observer for the state and the perturbation of the Stewart platform, which is combined with a variable structure controller (VSC). The combination of controller and observer provides the robust routine called sliding mode control with sliding perturbation observe. (SMCSPO). The optimal gains of SMCSPO, which is determined by nominal eigenvalues, are easily obtained by genetic algorithm. The proposed fitness function that evaluates the gain optimization is to put sliding function. The control performance of the proposed algorithm is evaluated by the simulation and experiment to apply to the Stewart platform. The results showed high accuracy and good performance.

• 한국정밀공학회지
• /
• 제21권4호
• /
• pp.120-131
• /
• 2004

In this paper, we propose the method of a motion planning generation in which the movement of the 3-link leg subsystem is constrained to a slider-link and a singular posture can be easily avoided. The proposed method is the jumping control moving in vertical direction which mimics a cat's behavior. That is, it is jumping toward wall and kicking it to get a higher-place. Considering the movement from the point of constraint mechanical system, the robotic system which realizes the motion changes its configuration according to the position and it has several phases such as; ⅰ) an one-leg phase, ⅱ) in an air-phase. In other words, the system is under nonholonomic constraint due to the reservation of its momentum. Especially, in an air-phase, we will use a control method using state transformation and linearization in order to control the landing posture. Also, an iterative learning control algorithm is applied in order to improve the robustness of the control. The simulation results for jumping control will illustrate the effectiveness of the proposed control method.

• 로봇학회논문지
• /
• 제2권4호
• /
• pp.297-303
• /
• 2007

In this paper, we discuss a service to give alarm in the case of emergency for the old and the weak by human behavior recognition in Intelligent Space. Our Intelligent Space consists of mobile robots, sensors and agents. And these components are connected to network framework. Agent analyzes data acquired from networked sensors and determines task of robots and a space to provide a service for humans. In our emergency alarm service, human behavior recognition service module analyzes accelerometer data obtained from body-attached human behavior sensing platform, and classifies into four basic human behavior such as walking, running, sitting and falling-down. For the old and the weak, falling-down behavior may bring about dangerous situations. On such an occasion, agent executes emergency alarm service immediately. And then a selected mobile robot approaches fallen person and sends images of the person to guardians. In this paper, we set up a scenario to verify the emergency alarm service in Intelligent Space, and show feasibility of the service from our simulation experiments.

• 산경연구논집
• /
• 제11권10호
• /
• pp.7-16
• /
• 2020

Purpose: It suggests that making a policy and strategies in a way of AI and its impact of commercialization on economic efficiency, social custom ethics. Research design, data, and methodology: The paper has analyzed the data based on the proposed model when derived as AI vs. FI job, etc. It is very different for each professional evaluation, which is artificial intelligence or robot job. One concept case was selected as a substitute job, with a relatively low level of occupation ability, such as direct labors, easily replaced. By the induction data has resulted in modeling. Results: The paper suggests that AI at high level become something how to make real decisions on ethical value modeling. Through physical simulation with the deduction data, it can be tuned to design and control what has not been solved, from human senses to climate. Conclusion: For the exploiting of new AI decision-making jobs in markets, the deduction data is possible to prove to AI's Decision-making that the percentage who can easily have different leadership as is different for each person. what is generated by some information silos may be applied to occupation societies. The empirical results indicate the deduction data that if AI determines ethical decisions (VC) for that modifications, it may replace future jobs.

• 한국항행학회논문지
• /
• 제17권4호
• /
• pp.386-395
• /
• 2013

본 논문에서는 AOA (Angle of Arrival)기반의 재머위치 결정기법에 사용되는 DOA 추정 알고리즘을 대상으로 Root-MUSIC으로 1차 DOA 추정을 수행한 후 MUSIC을 이용한 2차 DOA 추정을 수행하는 성능 개선 알고리즘을 제시하였다. 또한 제안한 알고리즘을 검증하기 위해 소프트웨어 기반의 시뮬레이션 플랫폼을 구성하여 오차환경에 따른 DOA 추정 성능 분석을 수행하고, 본 논문에서 제시한 DOA 추정 알고리즘의 DOA 추정 정확도 및 분해능의 개선 정도를 분석하였다.