VLA-Agent with flexible modality addition and removal capabilities
First author ・ 2025
Proposal of a Method for Integrating Modality Information Using Tool-Call-Type VLA Agents
About
I am a second-year student at Wakayama KOSEN.
ーJapan’s colleges of technology offering practical engineering education from age 15.
My research focuses on Physical AI, 4D, and VLM, with particular interest in achieving Spatial Intelligence and implementing it in Robotics and VR.
At cvpaper.challenge — Japan's largest AI research community — and at LIMIT Lab, our international collaborative research group, I am conducting research on data- and computation-efficient machine learning models under the supervision of Hirokatsu Kataoka.
Projects
First author ・ 2025
Proposal of a Method for Integrating Modality Information Using Tool-Call-Type VLA Agents
・ 2023
An icon generator app themed around Japanese traditional family crests (kamon)
First author ・ 2025~
In research
co-developer ・ 2025~
In deveplopment
Talks
cvpaper.challengeから始まる研究ライフ ・ 2024-07 ・ Team Lab Studio, Tokyo
Poster presentation ・ 2025 ・ Kyoto research park, Kyoto
Poster presentation ・ 2025 ・ Ritsumeikan University, Osaka