Toyota Research Institute (TRI) is developing general-purpose robots capable of accomplishing a wide variety of dexterous tasks by building general-purpose machine learning foundation models for dexterous robot manipulation, called Large Behavior Models (LBMs). These LBMs use generative AI techniques to produce robot action from sensor data and human requests, aiming to improve the quality of human life and amplify the human experience in mobility.
Requirements
- Hands-on experience with using machine learning for learned control, including RL, offline RL, or behavior cloning, for manipulation. Or, experience with machine learning and familiarity with large multi-modal datasets and models.
- Strong software development skills in Python.
- Experience deploying policies on robots in the real world.
- Experience working in a research environment.
- Experience training foundation models with large-scale datasets.
Responsibilities
- Learning from egocentric video: Using first-person human data to instruct robots how to act and reason about tasks in the real world.
- Pretraining with human data: pretraining Robot Foundation Models for Dexterous Manipulation using human data at scale
- Instruction following and grounding: Improving how LBM-powered robots interpret and act on natural language and multimodal commands.
- Interactive learning: Building robot agents that can engage with humans to reduce ambiguity in goals and instructions.
- In the wild data collection and learning: Developing scalable methods to acquire, filter, and learn from diverse, unstructured real-world data.
- Learning at scale: Designing data pipelines and model architectures that efficiently train on massive, heterogeneous datasets.
- Create working code prototypes, interact frequently with team members, run experiments with both simulated and real (physical) robots, and participate in publishing the work to peer-reviewed venues.
Other
- This is a summer 2026 paid 12-week internship opportunity.
- Please note that this internship will be an in-office role.
- A “make it happen” attitude and comfort with fast prototyping.
- A passion for robotics and doing research grounded in important fundamental problems.
- comfortable working with both existing large static datasets as well as a growing dynamic corpus of robot data.
