We are a startup focused on embodied intelligence applications, dedicated to combining AI, robot control, and real-world applications to build intelligent robotic systems with autonomous perception, decision-making, and action capabilities.
Based on your background and interests, you will primarily focus on either Locomotion or Manipulation, with opportunities to work on cross-domain topics such as mobile manipulation and full-body control for humanoid robots. Candidates can focus on either Locomotion or Manipulation; experience across motion control, manipulation control, and real-robot deployment is preferred.
Responsibilities
- Participate in the R&D of robot motion intelligence algorithms, including Locomotion, Manipulation, Whole-Body Control, and Motion Planning.
- Participate in motion control algorithm development for quadrupedal, bipedal, humanoid, or mobile robots, including gait generation, balance control, foot-end trajectory design, velocity tracking, and posture control.
- Participate in manipulation control algorithm development for robotic arms, dexterous hands, or mobile manipulators, including grasping, trajectory planning, end-effector control, impedance control, force control, and visual servoing.
- Participate in policy training, debugging, and evaluation using reinforcement learning, imitation learning, trajectory optimization, MPC, or traditional control methods.
- Build, train, and validate locomotion or manipulation policies in simulation environments, and participate in sim-to-real transfer.
- Participate in real-robot testing and analyze data such as joint states, IMU, force/torque, contact states, camera data, depth maps, and point clouds.
- Read and reproduce English papers and open-source projects in robot control, embodied intelligence, legged locomotion, robot manipulation, imitation learning, and reinforcement learning.
- Collaborate with algorithm, software, and hardware teams to integrate algorithm modules into robotic systems and improve system stability, real-time performance, and maintainability.
Requirements
- Major in robotics, automation, computer science, AI, mechanical engineering, electronics, control, or related fields; senior undergraduate, master's, or PhD candidates are preferred.
- Familiarity with at least one of the following areas: robot kinematics, dynamics, control theory, path planning, reinforcement learning, or imitation learning.
- Proficiency in Python with strong skills in algorithm implementation, experiment analysis, and data processing.
- Familiarity with C++ and ability to contribute to development in robot control, simulation, algorithm deployment, or system integration.
- Familiarity with Linux development environments and proficiency with Git for code management.
- Understanding of at least one robotics simulation platform, such as MuJoCo, Isaac Sim / Isaac Lab, Gazebo, PyBullet, or Webots.
- Understanding of at least one algorithm direction, such as Locomotion, Manipulation, Whole-Body Control, Motion Planning, MPC, RL, Imitation Learning, or Diffusion Policy.
- Strong mathematical foundation in linear algebra, optimization, probability and statistics, classical control, or robot dynamics.
- Ability to read English papers and technical documentation fluently, with strong self-learning and problem decomposition abilities.
- Strong interest in robotics and embodied intelligence, with willingness to engage in simulation debugging, real-robot testing, and complex engineering troubleshooting.
Nice to Have
- Project experience with quadruped robots, bipedal robots, humanoid robots, robotic arms, dexterous hands, mobile robots, or mobile manipulators.
- Experience in legged locomotion, humanoid locomotion, arm manipulation, dexterous manipulation, bimanual manipulation, or mobile manipulation.
- Familiarity with ROS / ROS2 and experience in robotic system integration or real-robot deployment.
- Familiarity with robotics dynamics, control, or planning tools such as Pinocchio, Drake, RBDL, Raisim, OCS2, CasADi, MoveIt, and cuRobo.
- Familiarity with reinforcement learning or deep learning frameworks such as PyTorch, JAX, Stable-Baselines3, rl-games, and rsl_rl.
- Sim-to-real experience and understanding of engineering issues such as control frequency, system latency, sensor noise, contact uncertainty, and dynamics randomization.
- Experience in robot teleoperation, data collection, imitation learning, behavior cloning, Diffusion Policy, ACT, or VLA.
- Experience combining vision, depth cameras, point clouds, force control, tactile sensing, and multimodal perception for manipulation.
- Experience in robotics competitions, research projects, open-source projects, publications, or demonstrable demos.
- Familiarity with AI coding tools and the ability to use them for paper reproduction, code development, and experiment analysis.
What You'll Gain
- Deep involvement in core locomotion and manipulation capability development for embodied intelligence robots.
- Exposure to the full robotics R&D workflow from algorithm design and simulation training to policy evaluation and deployment on real robots.
- Hands-on work in frontier directions such as Locomotion, Manipulation, mobile manipulation, and full-body control for humanoid robots.
- Close collaboration with algorithm, software, and hardware teams to solve complex challenges in real robotic systems.
- Top performers may receive full-time conversion or long-term collaboration opportunities.
