Publications

Selected papers and drafts

This page collects selected publications and current drafts spanning imitation learning, world models, task and motion planning, and robot foundation models. For the broader academic profile, see the CV page.

Publication Themes

  • Robot learning
  • Imitation learning
  • World models
  • Dataset scaling
  • Task and motion planning
  • Long-horizon manipulation

Recent papers and ongoing drafts

A focused list of papers and drafts relevant to robot learning and embodied intelligence. The CV contains the broader academic record.

Working Draft

SAIL: Test-Time Scaling for In-Context Imitation Learning with VLM

Makoto Sato, Yusuke Iwasawa, Yujin Tang, So Kuroki

arXiv draft, 2026

Conference Paper

Task and Motion Planning Using Residual Reinforcement Learning for Long-Horizon Precise Object Manipulation Task

Makoto Sato, Yuhwan Kwon, Yoshihisa Tsurumine, Takamitsu Matsubara

SCI, 2024

Conference Paper

Scaling Laws of Model Size for World Models

Makoto Sato and collaborators

JSAI, 2023

Conference Paper

Scaling Laws of Dataset Size for VideoGPT

Masahiro Negishi, Makoto Sato, and collaborators

JSAI, 2023

Conference Paper

Construction and Validation of Action-Conditioned VideoGPT

Koudai Tabata, Junnosuke Kamohara, Makoto Sato, and collaborators

JSAI, 2023

Conference Paper

Imitation Learning with Mid-Level Representations for Object Rearrangement

Makoto Sato and collaborators

JSAI, 2022

Research threads behind the papers

  • Scalable robot learning How larger demonstrations, richer representations, and better simulation change policy performance.
  • World models and scaling laws How model size and dataset size affect embodied generative models.
  • Planning and manipulation How task and motion planning can support long-horizon precision.

Need the full list?

  • CV PDF Download the latest curriculum vitae for a fuller publication and experience list.
  • Research page Browse the research overview for project context and themes.
  • Contact Reach out if you need details, slides, or discussion about related work.