Even though artificial intelligence (AI) technology has been significantly advanced in the past decade, we still have many problems to address in the fields of AI and robotics, including self-driving vehicles. Creating a long-term semiotic communication between AI and humans is a challenging problem towards involving AI technology in real world applications. Self-driving vehicles, for example, need to effectively cooperate with drivers so as to estimate the fatigue level of drivers, to offer them high-level technology services, etc. Home robots need to perform tasks at home environment in an autonomous manner, while communicating with the users and estimating their intentions.
Challenges to address in this project are:
１．Optimizing the system including both human and artificial intelligence to create a harmonized and human-centered system.
２．Developing a theory of machine learning methods that enables a robot to learn a variety of knowledge through daily human-robot interaction.
３．Developing a theory of semiotics that can model the dynamics of symbol systems.
In order to address these problems, we should cope with interdisciplinary academic challenges that span information science, engineering, and humanities through a next-generation artificial intelligence technology. To meet this target, we are creating an international research network that constitutes a basic motor for interdisciplinary research on artificial intelligence and semiotics.
Over and above, in order to develop such a next-generation artificial intelligence, we shed light on the possibility of business-academia collaboration and business opportunities related to AI development. In addition, demonstrating the impact of the developing technology on society considering the needs of aging people, we would proactively satisfy societal needs of technology during the next few decades. Researchers in this project are distributed over three campuses: BKC, Kinugasa and OIC, and synergistically collaborate to produce a remarkable outcome.
Goal and Challenges
The main goal of the project is to develop the "next-generation artificial intelligence" as a basis for integration between artificial intelligence and human users. Our target is not just developing a new industrial technology, but creating an academic community involving the related fields:
１．Machine learning and Artificial intelligence in Information Science and Engineering.
２．Self-driving cars and home robots in Science and Engineering.
３．Innovation in semiotics through continuous interaction between researchers related to AI and others related to humanities.
The group leader, professor Tadahiro Taniguchi, has been developing "Symbol Emergence Systems Theory" as an academic theory connecting information science and semiotics since 2010, defining an emerging field called "Symbol Emergence in Robotics (SER)". Recently, the academic community studying SER in Japan has been attracting attention, which we would try to enhance during this project through creating an interdisciplinary national and international research line focusing on symbol emergent system theory.
In order to evaluate the developed technology concretely, we will participate in several competitions including the RoboCup@Home League, and the World Robot Summit, which will be held in conjunction with the Tokyo Olympics (2020), so as to pave the way towards creating new real-world applications.
We will create "Symbol Emergence System Theory" as a interdisciplinary research field integrating artificial intelligence research (including automatic driving cars, domestic robots, machine learning methods, etc.) and semiotics. We will build a model that explicitly includes the cultural aspects of semiotic communication and embodied sensor-motor interaction between humans and between human and AI, give feedback to science, engineering and humanities, and form an interdisciplinary new academic field. This area is an extremely new area even from an international perspective.
New technologies (to be developed)
We will develop new technologies to help investigating the following problems:
・Artificial intelligence technology that dynamically resolves discrepancies in the intentions of a driver and a self-driving car.
・Artificial intelligence technology by which the driver's intention is estimated through the fusion of various sensor information using machine learning algorithms.
・Artificial intelligence technology that autonomously learns the function and usage of objects existing at home environment through a robot observing human behavior over time.
・Robots that could learn the maps of homes and offices, and, create a hierarchical place concept and object concept through interaction with human users so as to behave autonomously in space.
・Artificial intelligence technology that enable robots to acquire new vocabulary from human speech.
・Artificial intelligence technology that enable robots to clearly identify and adapt cultural differences between vocabularies and interpretations in different environments, e.g., homes and offices.
・Conceptual framework in semiotics for expressing and discussing symbol systems that dynamically change through interactions between agents.
・Creating new business opportunities inspired by the next-generation AI technology.
・Predicting the changing needs of aging people using the next-generation artificial intelligence.