Motion Control Engineer

Develop real-robot control systems, policy deployment interfaces, and stable execution loops.

What You Will Do

Motion Control System Development

Responsible for the R&D and deployment of robot motion control in embodied intelligence scenarios. Support stable execution, debugging, optimization, and continuous iteration on real robot systems such as robotic arms, dexterous hands, and mobile manipulation robots.

Low-Level Motion Control and Debugging

Develop and debug low-level motion control pipelines for robotic arms, including joint control, Cartesian control, end-effector pose control, trajectory tracking, gripper control, velocity/acceleration constraints, safety boundaries, and control frequency adaptation, ensuring stable, safe, and continuous motion.

Policy Deployment and Tuning

Work with algorithm teams to deploy and tune robotic arm policies on real machines. Analyze failure causes and optimize action space, control interfaces, execution frequency, action smoothing, trajectory constraints, safety strategies, and task workflows to improve real-robot success rate and stability.

Toolchain and Integration

Develop and maintain control interfaces, policy deployment interfaces, data collection tools, task scripts, automated evaluation tools, replay analysis tools, and experiment management workflows, improving the efficiency of real-robot experiments and policy iteration.

End-to-End Model Deployment

Integrate and debug end-to-end manipulation models such as ACT, Diffusion Policy, OpenVLA, and π0. Build the connection from model outputs to robot control execution, resolving adaptation issues between model actions and low-level control interfaces.

Real-Robot Feedback Loop

Establish a closed loop of “real-robot execution → data collection → failure analysis → control/policy adjustment → redeployment validation.” Continuously optimize robot manipulation policies using human feedback, model feedback, task success rates, and failure samples.

Stability Optimization

Troubleshoot and optimize issues such as jitter, discontinuous trajectories, excessive latency, frequency mismatch, collision risk, grasping failure, end-effector drift, and actuator anomalies, improving robustness in complex tasks.

Cross-Team Collaboration

Work closely with algorithm, data, and hardware teams. Understand the requirements of embodied intelligence models for robot control execution and transform algorithm outputs into action capabilities that can be stably executed by real robots.

What We Expect

Academic Background

Background in robotics, control engineering, automation, mechatronics, computer science, artificial intelligence, electronic information, or related fields. Bachelor’s degree or above, with a solid foundation in robot kinematics, control theory, or robot system development.

Robotic Arm Motion Control Ability

Familiar with low-level robotic arm motion control processes. Understands joint control, end-effector pose control, trajectory planning, trajectory tracking, velocity/acceleration constraints, impedance control, force control, and gripper control. Real robotic arm debugging experience is preferred.

Real-Robot System Experience

Experience in real robot system development, debugging, or deployment. Able to handle issues such as control instability, motion jitter, trajectory anomalies, communication latency, actuator anomalies, safety protection, and task failures during robot operation.

Policy Tuning Collaboration Ability

Able to work with algorithm teams on real-robot tuning of robotic arm policies. Understands the relationship between model output actions and low-level control execution, and can help improve real-robot policy performance from perspectives such as control interfaces, action smoothing, execution frequency, observation latency, coordinate transformation, and safety constraints.

Programming and Engineering Implementation Ability

Proficient in at least one language among Python/C++. Familiar with common development tools such as Linux, Git, and Docker. Strong engineering implementation skills, able to develop control interfaces, debugging scripts, data collection tools, evaluation tools, and deployment tools.

Robot Software Stack Experience

Familiar with ROS/ROS2 or other robot middleware. Understands robot communication, TF coordinate transformations, sensor integration, control nodes, motion planning interfaces, and task execution workflows. Experience with MoveIt, Pinocchio, Drake, Mujoco, Isaac Sim/Lab, etc. is preferred.

Understanding of End-to-End Models

Understands the basic paradigms of robot end-to-end manipulation models, VLA models, or imitation learning methods, such as ACT, Diffusion Policy, OpenVLA, and π0. Understands the deployment requirements and constraints of algorithmic models in real robot control pipelines.

Real-Robot Closed-Loop Awareness

Understands real robot data collection, task evaluation, failure case analysis, hard-case mining, and policy iteration workflows. Able to participate in building policy deployment and optimization loops for real-world tasks.

System Troubleshooting Ability

Strong problem localization and engineering debugging ability. Able to analyze real-robot problems across multiple levels, including algorithm outputs, control interfaces, coordinate systems, temporal synchronization, sensors, actuators, communication links, and safety mechanisms.

Collaboration and Implementation Ability

Strong hands-on ability, sense of responsibility, and cross-team collaboration skills. Able to work deeply with real robot systems and collaborate with algorithm, hardware, electrical control, mechanical, and data teams to drive the deployment of embodied intelligence capabilities.