As cell phones have become more common, personalized intelligent services in smartphones have become more highly desired. The mobile intelligent synthetic character is an example of one of these desired services. It is hard to apply an intelligent synthetic character to the smartphone environment because of its dynamism and complexity. This paper proposes a method for generating behaviors of a smart synthetic character that infers user contexts with Bayesian networks. In order to generate more realistic behaviors, the OCC model is utilized to create the character’s emotion. Behaviors are produced through large-scale modular behavior networks with inferred contexts. A working progress is the mobile log collected with a Samsung SPH-M4650 smartphone that is used to verify the naturalness and flexibility of the generated behaviors.
Highlights
► The proposed mobile intelligent synthetic character works out the problems which can occur in the mobile environment. ► It integrates Bayesian networks, the OCC model and the behavior networks. ► It consists of five components: a perception system, a user context recognition system, a character emotion system, a motivation system and a behavior generation system. ► To develop the proposed intelligent synthetic character, we used Microsoft Embedded Visual C++ 4.0 and Pocket PC SDK 2003. ► The Wilcoxon signed-rank test confirms that the proposed method was more appropriate.
High-end devices on mobile networks, smartphones, have been developed to include both functions of a PDA (personal digital assistant) and those of a cell phone. As smartphones have become more common, demand for personalized intelligent services will increase. The intelligent synthetic character we focus on is an autonomous agent that can interact with a user in real-time. The character provides information to users or interacts with them based on their facial expressions, behaviors, and simple dialogs. The intelligent synthetic character can also be used by entertainment and service robots (Kim, Kim, Kim, & Lim, 2002). Many researchers have attempted to apply this intelligent synthetic character to smartphones.
The main issue when implementing a synthetic character is how to make the character realistic and believable. Many researchers have aimed to designate character that seems to think, feel and live. To achieve this goal, the characters should have their own perceptional, behavioral and motor mechanisms (Pan, Yang, Xu, & Zhang, 2005).
The intelligent synthetic character should provide suitable services for various situations. To this end, there are three general required functions of an intelligent synthetic character. First of all, the character should recognize the contexts around the user. Not only low-level contexts gained directly from sensors but also high-level contexts inferred from various low-level contexts can be used. Secondly, the character should generate suitable behaviors for various situations by considering the inferred contexts. Moreover, to make the character more believable, it should behave according to its own internal state, since humans have a wide variety of expressive actions in their personalities, emotions, and communicational needs (Canamero, 1998 and Cassell, 1999).
However, since the mobile environment is often complex and changes dynamically, the mobile intelligent synthetic character should be able to handle the mobile environment and its structure. For example, the character should produce reliable information and be aware of contexts in the presence of uncertain, rapidly changing, and partially true data from multiple heterogeneous sources (Korpipaa, Mantyjarvi, Kela, Keranen, & Malm, 2003). The character should also generate behaviors as fast as possible in real-time by considering recognized contexts, since the mobile environment changes rapidly. In this paper, we propose a mobile intelligent synthetic character that considers problems which can occur in the mobile environment.
The proposed mobile intelligent synthetic character was designed to integrate Bayesian networks, the OCC model and the behavior network. To solve the problem of context inference in the mobile environment, we used a Bayesian network, are presentative probabilistic approach. Also, the behavior network was used to quickly generate behaviors suitable for various situations. Finally, the OCC model was used to generate the emotional state of the character in order to make it more realistic.
We proposed a mobile intelligent synthetic character which can deal with dynamism and complexity in a mobile environment. A Bayesian network was used to infer high-level contexts by considering the uncertainties of the sources from the smartphone, and the OCC model was used to generate the character’s emotional state. To provide numerous behaviors which can be performed in various situations by considering inferred situations and the character’s internal states, we proposed a large-scale behavior network with many behaviors.
In order to provide enhanced intelligent services for smarter phones, it is necessary for the intelligent synthetic character to interact with the user and to evolve independently. The character should be realistic. To achieve this, we developed algorithms for interaction and evolution, including a learning system that evolves the structure of the Bayesian networks and the behavior selection network using user feedback.
