Developing data-driven/ML-powered robotic control policies for dextrous manipulation and teleoperation applications at Facebook Reality Labs
Requirements
- Experience with RoS, Python, PyTorch or JAX, or other related languages
- Understand state of the art in robotic manipulation control policies, and advancing the technological frontier in an impactful and efficient way
- Experience with modern control policies like reinforcement learning, imitation learning, and large behavioral models
- Familiarity with modeling and analysis used in robotics including kinematics, dynamics, motion planning, perception, task planning, and control theory
- Experience working with robot manipulation
- Experience with multimodal self supervised representation learning
- Experience on vision based input recognition systems, such as hand tracking and pose estimation
Responsibilities
- Conduct collaborative research on developing control policies for a range of robotics platforms.
- Implement frameworks to train state-of-the-art machine learning control policies such as Large Behavioral Models, Imitation Learning, Reinforcement Learning and VLAs.
- Develop, implement, and evaluate methods for learning robust representations from multi-modal data (e.g., video, audio, IMU, EMG, Tactile).
- Develop robotic data collection pipelines for robotic dextrous manipulation, train models, and benchmark different algorithms.
- Collaborate with researchers and cross-functional partners including communicating research plans, progress, and results.
Other
- Currently has or is in the process of obtaining a Ph.D. degree in Computer Science, Artificial Intelligence, Robotics, or relevant technical field
- Must obtain work authorization in the country of employment at the time of hire and maintain ongoing work authorization during employment
- Intent to return to the degree program after the completion of the internship/co-op
- Availability for minimum 16 consecutive week internship
- Proven track record of achieving significant results as demonstrated by grants, fellowships, patents, as well as first-authored publications at leading workshops or conferences
