About研究室概要

Human-agent interaction (HAI) has become an important field in the field of human-computer interaction (HCI). The agent in HAI behaves with users as if it has its own intentions. It triggers users’ social responses, and instructs users through social channels. The use of HAI is widespread from the field of entertainment to medical purposes. The robots Aibo and Pleo are commercialized pioneers in HAI that entertain users with their behaviors by changing user’s internal states. The seal-like robot Paro, AuRoRA projects, and other agents have tried to solve dementia and autism through their behaviors. Virtual agents have established intimate relationships with users through their behaviors, and they have even accelerated sightseeing in the real world.

The term agent means an acting mediator. The use of the term has been extended in several areas in human-computer interactions, and there are many definitions for agents. We defined an agent as an artificial social actor that is accepted by users through her/his intentional stance, based on Dennett’s Intentional Stance, whether users are conscious or unconscious of the fact. The definition can include the unconscious social behaviors of users presented in media equation studies. The definition also includes a suspension of disbelief in human-agent interactions as explained by Duffy et al. where people accept the system as an intentional agent consciously to reduce cognitive load, even if they know that this is really not like watching drama on a stage.

The goal of our lab is to proposing several usage of agent and agency to the society. Human-agent interaction is on the intersection of artificial intelligence, human interface. cognitive science and other related studies. Everything related to the HAI is our target. For example, what kind of expression evoke a user as an agent (like shape, motion, behavior, and auditory and visual changes), how to create credible agent, and the mutual behaviors of agents and effect for users.

自律して動くロボットや、音声・スクリーン上のキャラクターを使用する情報技術が社会の中に増えてきました。これらの自律したシステムの振る舞いを観察すると、人間はそのシステムをただの道具としてではなく「擬人化」して捉え、他者に対するのと同様の社会的やり取りを、人工的システムに対し意識的あるいは無意識的に期待するようになります。
このように、人間から社会的なやり取りを期待される人工システムを、我々の研究室では道具(ツール)と対比して「エージェント」と呼んでいます。
人の間に介在する社会的存在であるエージェントが、人間にどのように働きかけられるかを検討し、使用場面に合わせたエージェントをユーザに対し提示することで、人工システムはユーザの行動を予測し、認識し、ひいてはユーザに対する誘導・教示を行うことが可能となります。

本研究室では、人間と人間の認知したエージェントとのやり取りを、対人実験を介して分析すると同時に、センサ、アクチュエータ、プログラムを組み合わせ、人とシステムの間に介在する新しいエージェントシステムを作り出し、将来の社会の中で活躍する人工システムのあり方を探ります。