Fascinated by our dreams of understanding human intelligence and bringing such intelligence to life in a robot like the Honda humanoid ASIMO, our research in computer science currently centers on computational learning. We focus on three major fronts: perception, human cognition, and robotics.
With human perception, we study visual understanding of humans and their surroundings by developing new computer vision algorithms and unique sensing systems. In auditory processing of the human voice, we seek to understand human language in noisy environments and conversational situations. We consider both the real-time performance and audio visual acuity simultaneously in our research.
In human cognition, we are intrigued by our commonsense decision making and learning abilities to improve human machine interaction. Our first efforts have investigated knowledge capture and representation, decision process modeling, and real-time performance analysis. We intend to develop our learning just as humans do, over time, at a system level for our robots.
Our ultimate goal is to realize all these new innovations in our humanoid robot. ASIMO presents an exciting and challenging platform for researchers in computational intelligence. In the robotics area, we attempt the ultimate demonstration of our research on a physical humanoid robot - going beyond simulations. This real stress-test will drive our future research directions as we pinpoint what needs to be improved or created. Our approach involves both the analysis and synthesis of knowledge and motions. With in-depth research in human motion skills, we hope to develop new ways to express these abilities on a humanoid robot in a more natural manner within the complex environs of everyday life.