📅 20th October, 2025 | ⏰ 09:00 – 13:00
Recent advancements in foundation models have revolutionized various domains of artificial intelligence, enabling generative design, adaptive control, and efficient optimization. In robotics, these models present a promising pathway for automated design synthesis, real-world adaptation, and simulation-to-reality transfer, accelerating the development of innovative robotic systems.
This workshop will explore the role of foundation models in robotic design, focusing on key applications such as generative design for novel morphologies, physics-informed learning, reinforcement learning enhancements, and cross-domain knowledge transfer. We will discuss how pre-trained models, self-supervised learning, and multimodal representations can empower roboticists to develop more efficient, adaptable, and intelligent robots.
Participants will gain insights into:
How foundation models can generate and optimize robotic structures and controllers
The impact of large-scale pretraining and fine-tuning in robotic applications
Challenges in real-world deployment, including data efficiency and robustness
Case studies on soft robotics, autonomous systems, and industrial automation
The workshop will feature invited talks, lightning talks, posters, and interactive discussions with leading researchers in AI and robotics. By bridging AI-driven design methodologies with robotics, it aims to foster collaboration and inspire new research directions.
This workshop is intended for researchers, engineers, and students interested in applying foundation models to robotic systems. No prior experience with deep learning is required, though familiarity with robotics or AI will be beneficial.
The University of Osaka
National University of Singapore
Tsinghua University
TU Delft
New York University
TBD
The University of Tokyo
University of Liverpool
The University of Tokyo
EPFL/NYU
University of Liverpool
EPFL