A Device Abstraction Framework for the Robotic Mediator collaborating with Smart Environments Young-Ho Suh, Kang-Woo Lee Intelligent Agent Research Team Electronics and Telecommunication Research Institute Deajeon, Republic of Korea {yhsuh,kwlee}@etri.re.kr Eun-Sun Cho Dep. of Computer Science and Engineering Chungnam National University Daejeon, Republic of Korea eschough@cnu.ac.kr Abstract— A robotic mediator is a robot platform that mediates between the user and the smart environment system in order to capture user’s intent and provide proactive services by collaborating with available resources in the environment. In order for a robotic mediator to provide the user with proactive context-aware services, it must communicate and collaborate with diverse heterogeneous devices and systems in the smart environment. However, existing network robot software platforms are not flexible enough to effectively integrate both of the respective features specific to robots and environments. Moreover, they do not support high-level collaboration model that is crucial to enable mediating between the user and the smart environment. In order to address the problem, we proposed ICARS [10], integrated control architecture for the robotic mediator in smart environments. ICARS consists of three layers each of which provides a flexible communication/device model, an adaptive service model for the integrated robot control architecture and a behavior-based high-level collaboration model as a key feature respectively. In this paper, we present the detailed design, implementation, and experiments of ICARS mainly focusing on the communication/device model layer. The experimental results show that ICARS enables flexible integration of the diverse devices in the robot and environment, adaptive service provision for collaborative services and easy development of high-level collaborative applications. Keywords-smart environment; robotic mediator; software framework; robot control architecture; collaboration I. INTRODUCTION The goal of a smart space is to support and enhance the abilities of its occupants in executing tasks in their daily living. Hence, smart spaces must be able to both detect the current context in the environment and determine what actions to take based on this context information. Namely, there must be a context-aware mediator between the user and smart spaces in order to capture the users’ intents and provide proactive services by organizing available resources in the environments. According to the vision that Mark Weiser [1] suggested, it was often the case that the mediator is embedded seamlessly and invisible to the user in the early systems. However, according to an experiment, 80% of subjects feel uncomfortable in interacting with an invisible presence and without explicit control over services [2]. Therefore, many projects have attempted interface technologies such as smart devices, wearable devices, tangible interfaces, spatial augmented reality, etc. to find an effective mediator for smart spaces, but none of them have been embraced to be killer applications yet [3][4][5]. Recently, since the concept of affective robots as mediators in smart spaces was proposed by a few research projects [3][6][7][8], the robotic mediators have attracted attention as a novel alternative to the new mediating interface. In order for a robotic mediator to provide the user with proactive context-aware services, it must communicate and collaborate with diverse heterogeneous devices and systems in the smart space [9]. However, it is not so straightforward to develop a robotic mediator since it is a complex distributed system consisting of a number of integrated hardware and software components. This requires a well-designed software framework which would interweave devices and services in an efficient manner to tackle various functionalities and requirements of developing robotic mediator systems. Fig. 1. The three abstraction layers of ICARS In [10], we proposed a novel software framework, i.e. ICARS which consists of four abstraction models as shown in Figure 1. We presented detailed design, implementation and experimental results of ICARS in the previous work. However, due to the lack of space, we described the overall architecture mainly focusing on the two upper layers of ICARS – collaboration model and control architecture. Thus, in this paper, we present the more detail description of the other two abstraction models – Device Abstraction layer and Communication Model. Typical software frameworks for the robotic mediator commonly provide the two abstraction models. The Communication Model is to hide low-level protocols and provides uniform communication interfaces. The Device Abstraction layer aims at encapsulating low-level device- specific details and provides high-level device models. However, most existing frameworks adopted general purpose middleware such as CORBA [11], UPnP [12], and SOAP [13], which results in suffering from inefficiency of performance and functionalities in exploiting robot-specific 2013 IEEE 16th International Conference on Computational Science and Engineering 978-0-7695-5096-1/13 $31.00 © 2013 IEEE DOI 10.1109/CSE.2013.75 460 2013 IEEE 16th International Conference on Computational Science and Engineering 978-0-7695-5096-1/13 $31.00 © 2013 IEEE DOI 10.1109/CSE.2013.75 460