A component is a reusable and replaceable software module accessed through its interface. Component‐based development is expected to shorten the development period, reduce maintenance costs, and improve program reusability and the interoperability of components. This paper proposes a new robot software component platform in order to support the entire process of robot software development. It consists of specifications of a component model, component authoring tool, component composer, and component execution engine. To show its feasibility, this paper presents the analysis results of the component's communication overhead, a comparison with other robotic software platforms, and applications in commercial robots.

Robot software components, which can be accessed through their interface, are reusable and replaceable software modules used in composing robotic services. Using robot software components in developing a robot, developers can expect shortening of developing times, reduction of maintenance cost and improvement of program reusability, because interoperability among components is guaranteed. We have developed a component framework assisting robot developers to make robot application with ease. The component framework consists of a specification of component model, a component authoring tool, a component composer, and a component container. In this paper, we will not only show the developed component framework, but also demonstrate its usability by applying to commercial robots.

Summary. This paper presents a new method of cooperative localization for multiple robots utilizing correlation between GPS errors of common mode in shared workspace. Assuming that GPS data of individual robot are correlated strongly as the distance between robots are close, we utilize the differential position data between the robots to refine robot's position data. Under artificial environment for simulation with imposed model error to robot motion and GPS sensor data error, it is confirmed that the proposed method provides improved localization accuracy [9]. In addition, we present a practical solution to accumulated position error in traveling long distance. IntroductionMobile robots require capability to estimate their position in order to navi gate autonomously in their work space. Consequently, localization by sen sor-based method has been researched as one of the most essential prob lems in mobile robotics. In previous researches, a number of works on lo calization of single robot have been reported [1,2]. However recently, many robotic applications require that robots work in collaboration in common workspace to perform a task. In such tasks as multiple robots op erate in close, we need more precise absolute localization or relative local ization of multiple robots in order to avoid collisions with each other. The multiple robot system, in comparison with the single robot system, has the advantages of collecting and integrating multiple sensor data from dif ferent robots. Accordingly, the system can obtain better localization per formance and increase the robustness of the localization accuracy for each

No abstract

Intelligent service robots leading the future market are robots which assist humans physically, mentally, and emotionally. Since intelligent service robots operate in a tightly coupled manner with humans, their safe operation should be an inevitable consideration. For this safety, real-time capabilities are necessary to execute certain services periodically. Currently, most robot components are being developed based on Windows for the sake of development convenience. However, since Windows does not support real-time, there is no option but to use expensive third-party software such as RTX and INTime. Also since most robot components are usually execute in user-level, we need to research how to support real-time in user-level. In this paper, we design and implement how to support real-time for components running in user-level on Windows using RTiK which actually supports real-time in kernel level on Windows.