This thesis proposes a novel control architecture for mobile printer system based on multi-robot cooperation. Mobile printer system consists of 3 omni-directional robot, wireless LAN system, GUI and 1 host computer. If users draw a picture in GUI, a host computer commands client robots to draw it on papers or something in the shortest time. To minimize elapsed time in drawing all lines, control architecture I, II, III are proposed. After client robots receive line look-up table from a host computer, they draw lines allocated to themselves in multi-robot control architecture II, III or they seek for the nearest line, draw it and repeat this process until all lines are drawn in multi-robot control architecture I. Control architecture I is a decentralized system and is based on behavior-based robotics. In this architecture, no planning is needed and mobile printer system has fault tolerance. On the contrary, control architecture II, III are centralized systems and are based on evolutionary programming. In control architecture II, III, a host computer finds time-optimal paths of all robots by EP and send them to all robots.
To solve several problems as well as elapsed-time optimization, proposed architecture is simulated and experimented. Simulation and experiment demonstrate the effectiveness and applicability of proposed multi-robot control architectures.
본 논문에서는 다개체 시스템을 이용한 모바일 프린터 시스템이라는 새로운 개념을 제안하고, 각 개체에게 작업량을 할당하는 최적화 제어구조를 제안하고자 한다. 전체 시스템은 한 대의 호스트 컴퓨터와 세대의 전방향(omni-directional) 동종(homogeneous) 로봇들로 이루어져 있다. 사용자가 호스트 컴퓨터에서 GUI(Graphic User Interface)를 통해 여러 개의 직선으로 이루어진 글씨나 그림을 그리고 나면 호스트 컴퓨터는 행동기반방식, 진화연산을 이용한 방식, 진화연산과 경험적인 방법이 결합된 방식의 세 가지 방식을 이용하여 로봇들에게 나누어줄 작업량을 최적화 하게된다. 최적화 된 작업량들을 무선랜을 통해 전달받은 로봇들은 최소의 시간 내에 글씨나 그림을 그리게 된다.