Recently, demand for service robots increases, and, particularly, one for personal service robots, which requires robot intelligence, will be expected to increase more. Accordingly, studies on intelligent robots are spreading all over the world. In this situation, we attempt to realize context-awareness for home robot while previous robot research focused on image processing, control and low-level context recognition. This paper uses probabilistic modeling for service robots to provide users with high-level context-aware services required in home environment, and proposes a systematic modeling approach for modeling a number of Bayesian networks. The proposed approach supplements uncertain sensor input using Bayesian network modeling and enhances the efficiency in modeling and reasoning processes using modular design based on domain knowledge. We verify the proposed method is useful as measuring the performance of context-aware module and conducting subjective test.
Recent robot researches have shifted the attention from industrial robots to service robots. According to the report of Japan Robotic Association (Fig. 1), robot market with personal and service robots as the center will grow exponentially in the near future. Since personal and service robots are investigated and made in order to provide services to individual users, they should have intelligence for various goals (Garcia, Jimenez, Santos, & Armada, 2007). For this reason, much research for robot intelligence appears from all over the world.
Full-size image (34 K)
Fig. 1.
Estimation of robot market in Japan.
Source: Japan Robotic Association.
Figure options
European Union investigated 73 million Euros in 74 projects with the topic of “Cognitive Systems and Robotics,” and it covers various topics of intelligent robotics including object recognition, cognitive architecture, activity modeling, planning, learning and adaptation (European Commission ICT Research). Researches on robotics in United States tend to seek practicality, and they include robot intelligence. Robotics Institute in CMU have studied service robots for people with reduced functional capabilities due to aging or disability (Brose et al., 2010), and personal robots group in MIT Media Lab have studied robots that can express their emotion and learn social interaction (Breazeal, Gray, & Berlin, 2009). Japan is in a dominant position in personal service robot field. In Japan, major companies such as Sony, Honda, and NEC as well as a government conducted related research projects. Sony’s AIBO and Honda’s ASIMO are well-known to the public as well as robot researchers, and they have been provided to major research groups as platforms for robot intelligence (Hing et al., 2008).
Despite much effort to implement robot intelligence, most of them focused on low-level control and recognition problems. We construct probabilistic model so that service robots in home environment can provide context-aware services to users and use modular design approach based on domain knowledge for efficient modeling. Bayesian networks are used to handle uncertain and various input values reliably. Bayesian network models are modularized and designed based on services and functionalities for efficient modeling and reasoning processes, and each module with functional independence can be used together if necessary.
The rest of this paper is organized as follows: Section 2 describes the backgrounds and related works on context-aware services in three different domains and general Bayesian network modeling, and Section 3 presents Bayesian network modeling approach based on modular design and the case study of context-aware home robot services. Section 4 evaluates the proposed modeling method in terms of performance and user satisfaction, and Section 5 concludes this paper and discusses future works.
This paper used Bayesian network modeling for service robots to provide users high-level context-aware services they need in home environment, and proposed a systematic modeling approach for modular design of a number of Bayesian networks. The proposed approach exploited Bayesian network to supplement uncertain sensor input and selected modular design approach based on domain knowledge to strengthen efficiency in modeling and reasoning processes. In experiment, we verified the proposed method is useful as measuring a performance of Bayesian network modules designed together with the service scenarios in home environment. Finally, we confirmed that user satisfaction is high enough with the subjective test of SUS.
