Ririko - Kinoshita

The transition from idol to actress is notoriously difficult. For every success story, there are dozens who fail to shed the “pop star” label. Kinoshita managed this transition by focusing on haien (stage plays) before moving to television.

Citation
Kinoshita, R., & Suzuki, M. (2016, October). Understanding user intent from eye‑gaze and hand pose for collaborative robots (Poster). Proceedings of the 2016 IEEE International Conference on Human‑Robot Interaction (HRI), 453‑454. https://doi.org/10.1109/HRI.2016.7747372 ririko kinoshita

Summary
Explores a multimodal sensor fusion approach (eye‑tracking + skeletal tracking) to infer a human collaborator’s intent within 0.5 seconds, paving the way for proactive assistance. The transition from idol to actress is notoriously difficult

Keywords
Human intent inference, eye‑gaze, hand pose, multimodal fusion, HRI. Citation Kinoshita, R

Citation
Kinoshita, R., Tanaka, K., & Sugimoto, H. (2019). Efficient semantic segmentation on low‑power embedded devices for assistive robotics. IEEE Transactions on Cognitive and Developmental Systems, 11(4), 617‑627. https://doi.org/10.1109/TCDS.2019.2913125

Summary
A lightweight encoder‑decoder network (named SqueezeSeg‑K) runs at >30 fps on a Jetson‑TX2 while maintaining >78 % mean IoU on the NYU‑Depth V2 indoor dataset.

Keywords
Semantic segmentation, embedded vision, assistive robotics, low‑power inference, SqueezeSeg‑K.