Robotics Intern, World Foundation Models

Toyota Research Institute•Los Altos, CA

1d•Hybrid

About The Position

Toyota Research Institute’s World Foundation Models team in the Robotics division is seeking research interns for summer 2027 across areas including video generation, world modeling, world-action models, 4D reconstruction, learning from egocentric videos, multimodal foundation models, multi-view geometry, and data augmentation, with a primary focus on embodied applications. The team works on scientific challenges in deploying robots in unstructured environments, using data from varied sources and modalities to learn transferable priors grounded in real-world physical properties. Its mission is to develop foundational models that understand how the world works, predict possible future states, and adapt to new environments and circumstances. As a Research Intern, you will work with a multidisciplinary team to propose, conduct, and transfer research at the intersection of computer vision and robotics. You will use large-scale, multimodal data to train foundation models, pursue open research problems, work toward publications at leading academic venues, and test ideas in simulators and the real world. This is a paid 12-week, hybrid, in-office internship.

Requirements

Currently pursuing a Ph.D. in machine learning, robotics, or a related field.
Publication record or desire to publish at high-impact conferences or journals in relevant areas.
Passion for large-scale machine learning and computer vision challenges grounded in physical systems, especially robotics.
Proficiency in one or more coding languages and systems; Python, Unix, and a deep learning framework such as PyTorch are preferred.
Ability to collaborate with researchers and engineers to invent and develop research ideas.
Reliable, collaborative approach, with a willingness to think ambitiously, investigate deeply, and deliver with integrity.

Nice To Haves

Experience or research interests in video generation, world modeling, world-action models, 4D reconstruction, learning from egocentric videos, multimodal foundation models, multi-view geometry, or data augmentation.
Familiarity with Python, Unix, and a deep learning framework such as PyTorch.
Publications or interest in publishing at venues such as CoRL, ICLR, NeurIPS, CVPR, ICCV, ECCV, ICML, UAI, AISTATS, AAAI, TMLR, RSS, ICRA, IROS, or RA-L.

Responsibilities

Conduct computer vision research addressing open problems of high theoretical and practical value, and evaluate solutions on real-world benchmarks and systems, with a focus on robotics.
Advance the state of the art in visual systems for robotics.
Collaborate with research scientists and engineers across the World Foundation Models team, Robotics division, Toyota Research Institute, Toyota, and university partners.
Stay current with state-of-the-art machine learning ideas and software.
Present research results verbally and in writing at international conferences and internally, and contribute to the community through open-source work.