Internship - Controls

Internship - Controls Here at Humanoid, we believe in a future where robots amplify human potential. That’s why we’ve set out on a mission to build the world’s most capable, commercially-scalable, and safe humanoid robots. We’re bringing that mission to life with HMND‑01 Alpha - our rapidly developed humanoid platform now running in real industrial pilots - and we’re growing the team to take it even further. THE OPPORTUNITY We’re looking for interns who are curious, hands-on, and excited to work directly with robotic systems. This is an open-ended internship where you will design and train reinforcement learning policies that enable dynamic locomotion and loco-manipulation behaviors on real robots. Your work will focus on building scalable training pipelines, designing reward functions and environments, and improving sim-to-real transfer for reliable deployment on hardware. You will work closely with control and robotics engineers to integrate learned policies into the robot control stack, ensuring stable and robust behavior in real-world conditions. This is a full-time internship (5 days per week) over the summer (mid June - mid September), based in our London Paddington office, where you’ll contribute to real robotic systems from early on with guidance from experienced engineers. Duration: 12 weeks | Start date: June | Compensation: Competitive pay + we'll keep you fed (seriously, our breakfasts and lunches are good) WHAT YOU MIGHT WORK ON - Design and train reinforcement learning policies for humanoid robot control - Build scalable simulation and training pipelines (e.g., Isaac Lab, MuJoCo) - Design reward functions, observation spaces, and curricula for complex behaviors - Run and analyse existing policies - Identify issues, troubleshoot, and propose creative solutions - Document procedures and findings, helping shape the evolution of our humanoids WHAT WE’RE LOOKING FOR - Hands-on experience with PyTorch and training ML models. - Strong interest in reinforcement learning, machine learning and robotics - Experience writing code (e.g. Python, C++, or similar) - Comfortable working with hardware, experiments, and debugging - Ability to learn quickly and operate in a fast-paced environment - Problem-solving mindset and attention to detail - Clear communication and ability to work closely with a team HOW TO APPLY Complete the challenge below and submit your solution as a public GitHub repository — include a README with instructions to run your system, example outputs, and a short note on your design choices. You will be able to include your GitHub repository URL when you fill out the application form, alongside your name and CV. We’re not looking for standard solutions, we're looking for how you think. The strongest submissions are creative, original, and push beyond the obvious. INTERN CHALLENGE: 3D END-EFFECTOR TRACKING Build a system that learns to control a robotic arm to track a desired end-effector trajectory. The major task is to achieve accurate, smooth position tracking over time (not just reaching a point). Requirements - Use a standard simulated arm (e.g., Franka, UR5, etc.) - Track a time-varying Cartesian trajectory (e.g., circle, figure-eight, moving target) - Use reinforcement learning as a core component - Ensure smooth and stable motion (avoid jitter) - Introduce at least one source of uncertainty: - noise (state or action) - Unreachable positions - control delay - (Optional) Add orientation tracking along with position. What to Submit - Self-contained code + instructions to run - Example results (plots and videos of tracking) - Short note covering: - state, action, reward design - how you represent the trajectory - how you evaluate tracking performance What We Care About - Tracking accuracy over time - Smoothness and stability - Robustness to noise/mismatch - Clarity and simplicity of your approach Make something you’re proud of.