Launch Your Foundation Model Research at the First RIG Bootcamp
Foundational Behavior Models represent a paradigm shift in robotics, yet Germany has a critical research gap in this emerging field. KIT, with support of the KIT Center Information · Systems · Technologies (KCIST), is hosting the first RIG Bootcamp on Foundational Behavior Models from November 24–28, 2025, to catalyze collaborative research across the German robotics community.
Why This Matters
Foundational Behavior Models represent a paradigm shift in robotics. The goal is to create robot generalists, i.e. systems that can learn and generalize behaviors across diverse tasks and environments. By teaching robots to understand and respond to language, vision, and physical interactions, we're working toward machines that can adapt to real-world complexity.
Startups, BigTech, and governments worldwide are investing billions in this technology. Yet Germany faces a critical research gap in this field, threatening our technological sovereignty and economic competitiveness in an area that will reshape manufacturing, healthcare, and the robotics industry. This bootcamp catalyzes collaborative research across Germany's robotics community to address this gap and accelerate Germany's research in foundation models for robotics.
Not a Workshop. A Research Sprint.
Rather than a traditional workshop or summer school, this is designed as a research accelerator. This intensive one-week research sprint brings together PhD candidates who will make substantial progress on concrete research projects through method derivation, prototype implementation, and initial experiments.
Participants collaborate with peers from KIT and other leading German research labs, working alongside international expert tutors with access to state-of-the-art robot hardware, substantial compute resources, and state-of-the-art VLA models. The week includes tutorials on the foundation model landscape and dedicated research time with expert guidance.
Projects won't be completed during the week, but participants will establish solid foundations and return home with a concrete project on a clear path to publication at major 2026 conferences (CoRL, ICRA, RSS, NeurIPS), supported by ongoing collaboration with KCIST and the RIG network.
What You'll Gain
Days of Intensive Research
Research Opportunities
Resources & Expertise
Collaborate with peers from KIT and other leading German research labs. Access state-of-the-art robot hardware, substantial compute resources, and expert guidance from international tutors. Gain comprehensive understanding of the foundation model landscape: RT-2, OpenVLA, π₀, FLOWER, BEAST, and emerging approaches.
Publication-Ready Projects
Launch research positioned for publication at CoRL, RSS, ICRA, or NeurIPS 2026. Participants will make substantial progress during the bootcamp and complete the project at their home university with advisor support.
Support
Participants will receive robot hardware, in particular tabletop mounted single and bimanual manipulators. Furthermore participants will have access to substantial compute resources and expert mentorship from leading researchers.
What We're Looking For
Research-Ready PhD Candidates
We seek projects that combine your existing expertise with foundation models to advance capabilities. Bring your domain knowledge in control, perception, manipulation, planning, or any robotics subfield and explore how foundation models can advance your research or how your expertise can advance foundation models.
Requirements
Strong implementation skills: Prepared to derive methods, implement prototypes, and run rigorous experiments.
Concrete research direction: Demonstrate that you can start immediately with a project scoped for completion at your home university within weeks after the bootcamp.
Home advisor support: Your advisor must commit to supporting project completion and publication submission.
Methodological innovation: We seek projects that push capabilities forward, leveraging foundation models to advance your robotics domain OR bringing your expertise to enhance foundation models themselves. We are not looking for projects that simply apply existing VLAs to new tasks without advancing the underlying methods.
Participation Agreement
Upon acceptance, participants will sign a participation agreement ensuring that research conducted during the bootcamp will be open-sourced and published collaboratively, benefiting the wider RIG community.
Apply Now
Join us in accelerating Germany's research in foundation models for robotics. Together, we'll advance the field and show that European research can compete at the highest level.