Keisuke Kinoshita

Research Organization of Information and Systems

Papers

Total Citations

H-Index

About

Keisuke Kinoshita is a pioneering researcher in the intersection of robotics, computer vision, and bio-inspired mechanical design. His work primarily addresses the fundamental challenge of **robotic control under sensory constraints**, particularly when systems must operate with only **partial visual information**. In his most influential work, Kinoshita tackled the difficult problem of control tasks where available visual data is perpetually insufficient at any given moment, yet must be accumulated over time to achieve successful manipulation—a critical insight for real-world robotic applications in cluttered or occluded environments. Beyond vision-based control, Kinoshita has made notable contributions to **biologically valid humanoid robotics**, specifically in creating naturalistic jaw movements for talking robots. His 2005 work on learning from demonstration (LFD) for jaw motion synthesis represents an early and important step toward building phonetically and visually synchronized, trainable robot mouth systems. This research addresses the perceptual challenge of naturalness in human-robot interaction, laying groundwork for more expressive and believable social robots. While his citation counts (7, 5, and 4 respectively) reflect a focused, niche impact rather than broad recognition, Kinoshita’s work is valued for its principled approach to hard problems at the boundary of perception, control, and biomechanics—offering foundational ideas for researchers building robots that must see, act, and speak in human environments.