Toyota Research Institute (TRI) is developing general-purpose robots capable of performing a wide variety of dexterous tasks, aiming to improve the quality of human life by creating robots that can assist with household chores, aid the elderly, and enable people to focus on enjoyable activities. The core challenge is enabling robots to operate reliably in unstructured environments and answering the question of what is needed to create truly general-purpose robots with minimal human supervision.
Requirements
- Data-efficient and general algorithms for learning robust policies leveraging multiple sensing modalities: proprioception, images, force, and dense tactile sensing
- Scaling learning approaches to large-scale models trained on diverse sources of data including web-scale text, images, and video
- Quick and efficient improvement of learned policies
- Experience with machine learning and familiarity with large datasets and models
- Strong software development skills in Python
- Experience in C++ is very helpful, but not strictly required
- Hands-on experience with using machine learning for learned control, including behavior cloning and/or reinforcement learning, for manipulation
Responsibilities
- create working code prototypes
- interact frequently with team members
- run experiments with both simulated and real (physical) robots
- participate in publishing the work to peer-reviewed venues
- working with both existing large static datasets as well as a growing and dynamic corpus of robot data
- Developing and deploying learned policies and complex mobile manipulator embodiments, such as humanoid robots
- Scaling learning approaches to large-scale models trained on diverse sources of data including web-scale text, images, and video
Other
- Currently pursuing a degree (Ph.D., M.S.) in Robotics, Computer Science, Mechanical Engineering, or a related field.
- Publication record at top-tier robotics/ML conferences (RSS, CoRL, ICRA, NeurIPS, ICML, ICLR, CVPR, ICCV).
- Hardware experience is strongly preferred, especially toward deploying learned policies on real robotic systems.
- A “make it happen” attitude and comfort with fast prototyping and running informative experiments.
- A passion for robotics and doing research grounded in important fundamental problems.
