Job Description
We're building the systems that let an advanced humanoid robotics platform pick up a box, open a drawer and operate tools. As a Robotics Researcher in Manipulation, you'll develop the grasp planning, contact-rich control and learned task policies that power the robot's hands. You'll work across model-based control, imitation learning and reinforcement learning, with the bar set by whether it works on the physical robot in a real environment, not just in simulation. This role combines research depth with a relentless focus on shipping to hardware.
\n
\n
What you'll do
\n
- \n
- Research, develop and deploy manipulation policies for dexterous task execution
- Build grasp planning and contact-rich control pipelines that generalise across varied objects and environments
- Design and run data collection and teleoperation infrastructure to feed policy training at scale
- Train manipulation policies using imitation learning, reinforcement learning or hybrid approaches, iterating until they work in the real world
- Integrate manipulation with the perception stack and broader autonomy pipeline
- Systematically diagnose hardware failure modes and drive improvement
- Contribute to open-source manipulation research and tooling
\n
\n
\n
\n
\n
\n
\n
\n
\n
What you'll bring
\n
- \n
- Strong foundations in robotics, control theory and motion planning
- Hands-on experience building and deploying manipulation systems on real robotic platforms
- Strong Python and C++; experience with PyTorch or JAX
- A track record of taking manipulation research from prototype to hardware deployment
- Experience with data collection infrastructure and teleoperation for policy training
\n
\n
\n
\n
\n
\n
\n
Nice to have
\n
- \n
- Diffusion policies, transformer-based policy architectures, or large-scale foundation models for manipulation
- Prior work on dexterous or in-hand manipulation
- Contact-rich or deformable object manipulation
- Publications at RSS, ICRA, CoRL or equivalent
\n
\n
\n
\n
\n