What you will be doing
Developing algorithms, foundation models, and methods for robotic manipulation and/or loco-manipulation, for industrial, scientific, and household applications;
Integrating these methods into real-world robotic manipulation systems, including those consisting of collaborative robot arms, industrial robot arms, mobile manipulators, humanoids, and dexterous hands;
Contributing to multi-person research projects that require a diverse set of skills across the robotics and machine learning stack;
Engaging with the academic community through high-impact publications, conferences, workshops, and/or code releases;
Developing AI agents to accelerate research and engineering while ensuring correctness, reproducibility, and security.
What we need to see
Enrolled in a PhD program in Robotics, Machine Learning, Computer Science, Electrical Engineering, Mechanical Engineering, or a related field;
Knowledge of both the theory and practice of robotics and AI, with a strong interest in connecting your work to real-world robotics applications;
A demonstrated track record, with first-author work published in top robotics and AI conferences and journals such as RSS, CoRL, ICRA, IJRR, T-RO, Science Robotics, NeurIPS, ICML, ICLR, CVPR, ICCV, ECCV, and EMNLP;
Exceptional communication, collaboration, and interpersonal skills, with significant experience working on a team;
Exceptional programming skills in Python; in addition, familiarity with C++, CUDA, and Warp is a plus;
Fluency in modern deep learning frameworks such as PyTorch or JAX;
Experience in AI coding and research agents, including critical validation of their outputs;
Experience with robotics frameworks such as ROS2 and physics simulation frameworks such as Isaac Sim, Isaac Lab, Newton, and/or MuJoCo;
Comfort in working through the complexities of simulation and real-world robotics, including debugging physics simulators and renderers under rapid development; selecting, setting up, maintaining, and enhancing complex robotics hardware; debugging communication systems; and designing robust workflows for model training and evaluation.
The following research areas and applications are of particular interest:
Bimanual and dexterous manipulation
Mobile manipulation and humanoid loco-manipulation
Multimodal robot learning (e.g., vision, tactile, and force/torque sensing)
Simulation (classical and neural), sim-to-real, real-to-sim, and world models
Learning in the real world (e.g., imitation learning, RL)
Robot foundation models (e.g., VLAs, WAMs, world models): pre-training, mid-training and post-training (SFT, RL, etc.) using various data sources (e.g., simulation, teleoperation, egocentric data)
Agentic robotics
Robot learning theory
High-precision manipulation and reliable policy deployment
Scientific lab automation