Papers
74
Total Citations
964
H-Index
17
About
Kazunori Komatani is a pioneering researcher at the intersection of robotics, machine learning, and human-robot interaction, with particular expertise in robot audition, neural network-based sensorimotor learning, and multimodal communication. His foundational contributions to robot audition — including the development of the HARK open-source software platform — have significantly advanced robots' ability to localize, separate, and recognize simultaneous speech in real-world acoustic environments, addressing one of the most challenging problems in embodied AI. Komatani's work on recurrent neural networks with parametric bias (RNNPB) has opened compelling pathways for robots to learn inter-modal mappings between language, motion, and visual perception through experience, enabling capabilities such as imitation learning and object dynamics prediction. His innovative research on musical robots — including systems that synchronize with live human musicians in real time — demonstrates a rare creative breadth that bridges technical rigor with human-centered applications. With papers accumulating citations across robotics, speech processing, and cognitive systems communities, Komatani's body of work, totaling over 387 citations across his top publications, reflects sustained and wide-reaching influence in building robots that perceive, communicate, and learn naturally within human environments.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4
- 5Making a robot recognize three simultaneous sentences in real-time33 citations · 2005
- 6
- 7
- 8Computational auditory scene analysis and its application to robot audition31 citations · 2004
- 9Inter-modality mapping in robot with recurrent neural network29 citations · 2010
- 10Experience-based imitation using RNNPB29 citations · 2007