Student Activities | Takato Horii | Osaka University

Four Presentations at JSAI 2026 (40th Annual Conference of JSAI)

Wed, 27 May 2026 00:00:00 +0000

Our lab will give the following four presentations at the 40th Annual Conference of the Japanese Society for Artificial Intelligence (JSAI 2026), to be held June 8–12, 2026 at G-Messe Gunma and online.

Organized Session: Physical AI in the Era of Foundation Models

Shogo Yanagida (Osaka Univ.), Tatsuya Aoki (Osaka Univ.), Tadahiro Taniguchi (Kyoto Univ. / Ritsumeikan Univ.), Takato Horii (Osaka Univ. / IRCN, Univ. of Tokyo) Tuesday, June 9, 2026, 13:30–15:00 — Room G (Main Hall A)

Sizhe Li (Osaka Univ.), Tatsuya Aoki (Osaka Univ.), Takato Horii (Osaka Univ. / IRCN, Univ. of Tokyo) Tuesday, June 9, 2026, 17:30–19:00 — Room G (Main Hall A)

Organized Session: AI Alignment

Kentaro Nomura (Osaka Univ.), Takato Horii (Osaka Univ. / IRCN, Univ. of Tokyo) Tuesday, June 9, 2026, 9:00–10:30 — Room I (Medium Conference Room 202A)

Organized Session: AI and Creativity — Beyond Imitation of Humans

Makoto Teshirogi (Osaka Univ.), Futa Hidaka (Osaka Univ.), Yuichiro Yoshikawa (Osaka Univ.), Takato Horii (Osaka Univ. / IRCN, Univ. of Tokyo) Thursday, June 11, 2026, 9:00–10:30 — Room G (Main Hall A)

For details, see the .

New Paper Published in IEEE Access: Distance-Aware World Model-based Reinforcement Learning

Thu, 21 May 2026 00:00:00 +0000

Our paper “Distance-Aware World Model-based Reinforcement Learning for Mobile Manipulation Behaviors,” authored by Xiaoxu Feng (doctoral student) and Takato Horii, has been accepted and published as Early Access (open access, CC-BY) in IEEE Access.

This work tackles end-to-end learning of mobile manipulation by jointly addressing embodiment selection (whether to move the base or use the arm) and motion planning within a single world model-based reinforcement learning framework. To overcome the limitations of similarity-based rewards in latent space, we explicitly incorporate 3D spatial information into the world model and derive a distance-aware reward for policy training. At the high level, a reachability-based reward enables rational switching between locomotion and manipulation. Extensive simulation experiments with both fixed and randomized target settings show that our approach achieves near-perfect motion success rates and substantially more rational arm activations than baseline composite-reward formulations.

This work was supported by JST Moonshot R&D (JPMJMS2011).

Authors: Xiaoxu Feng, Takato Horii (Graduate School of Engineering Science, The University of Osaka)
Journal: IEEE Access (Early Access, May 21, 2026)
DOI:
Paper:

Moe Nakamura Wins the Outstanding Presentation Award at JSKE Spring 2026

Wed, 15 Apr 2026 00:00:00 +0000

At the 21st Spring Conference of the Japan Society of Kansei Engineering (JSKE), held from March 16 to 18, 2026 at the Yoto Campus of Utsunomiya University, Ms. Moe Nakamura (POLA Chemical Industries; joining the lab as a working-adult doctoral student (D1) from April 2026) received the Outstanding Presentation Award.

Presentation title: “Contrastive Learning of Tactile Impressions and Physical Features from Free-Description Data – Connecting Free Verbal Expressions of Touch to Design”
Authors: Moe Nakamura, Kentaro Nomura, Koji Mizukoshi, Takaya Oishi, Toshiki Ikejima, Takato Horii

For details, see the .

Kentaro Nomura Selected as MVP at CPC Spring Camp 2026

Wed, 01 Apr 2026 00:00:00 +0000

At “CPC Spring Camp 2026,” a spring research camp on Collective Predictive Coding and Symbol Emergence Systems held from March 21 to 26, 2026, Mr. Kentaro Nomura (M2 in our lab) was selected as MVP.

For details, see the .