Dexterous Hand and End-Effector Control Intern

Job Responsibilities

• Assist engineers with design, code development, on-robot debugging, and performance optimization for control modules used by robotic dexterous hands, grippers, and other end effectors.
• Participate in the development and debugging of control logic and end-effector action strategies related to multi-degree-of-freedom movements of dexterous hands, finger coordination, stable grasping, and fingertip contact.
• Assist in completing interface adaptation, coordinate system calibration, kinematic modeling, and control integration for dexterous hands, grippers, long-handled tools, and soft-padded end-effectors.
• Participate in designing end-effector control strategies for robotic tasks such as grasping, clamping, pressing, flipping, rotating, pushing and pulling, lifting, and repositioning.
• Assist in building closed-loop control logic based on visual, force, tactile, current, and joint state feedback, improving the compliance, stability, and task success rate of end-effector operations.
• Debug the collaborative control logic of the robotic arm and the dexterous hand, and establish a complete operational chain from the robotic arm's pose control to the fine movements of the fingers.
• Participate in building bimanual coordination capabilities, supporting composite tasks such as bimanual grasping and handling, holding an object with one hand while performing fine operations with the other, and lifting and stabilizing with both hands.
• Assist in developing end-of-line operation safety strategies, including force limits, speed limits, collision protection, slippage detection, clamping stability assessment, and abnormality recovery logic.
• Participate in on-robot end-effector debugging and control-parameter tuning, review manipulation failures, and continuously improve end-effector control strategies.
• Conduct end-effector operation validation separately in simulation environments and on physical robots, promoting the implementation of tasks such as grasping, door opening, button pressing, and tool operation from prototypes to real-world scenarios.

1. Basic General Ability (Base)

• Proficient in Linux/Ubuntu development environments; experience with robot or robotic-arm simulation or on-robot debugging is preferred.
• Proficient in C and Python, with the basic ability to independently write and debug code, as well as implement projects.
• Familiar with the overall architecture of ROS/ROS2, understanding commonly used mechanisms such as Topic, Service, Action, TF coordinate transformation, parameter management, and rosbag.
• Master the basics of robot kinematics: forward and inverse kinematics, Jacobian matrix, coordinate transformation, SE(3), quaternions, pose calculation;
• Understanding the relationship between the robotic arm and the end tool coordinates can help complete TCP calibration, hand-eye calibration, and end tool coordinate system adaptation.
• Understand various basic robotic control methods such as position, velocity, impedance, admittance, and force–position hybrid.
• Has hands-on experience with robotic arms, dexterous hands, and grippers, and can assist in troubleshooting engineering issues such as grip slippage, force control oscillation, motion over-limit, and unstable contact.
• Proficient in using Git, with awareness of standard coding, documentation organization, testing and validation, and cross-module collaboration.

2. Specialized Technical Directions (meeting one or more is sufficient)

Track A: Dexterous-Hand Control and End-Effector Development

• Familiar with the hardware control architecture of multi-fingered dexterous hands and electric grippers; knowledgeable about theories related to finger coordination, joint limits, grasp constraints, and grasp stability; able to assist in the development of dexterous hand drive interface encapsulation, action command design, and state feedback analysis; capable of designing finger coordination action logic for tasks such as grasping and pressing; candidates with experience in dexterous hands, tactile gloves, or custom end-effector tool debugging are preferred.

Track B: Grasping, Contact, and Fine-Motion Control

• Understand the basic concepts of robot grasp planning, grip force control, and slip detection; be able to use vision, point cloud, and depth information to assist in planning grasp points and finger closure strategies; understand differentiated handling methods for soft and hard, large and small, irregular objects; be familiar with fine operation scenarios such as buttons, knobs, door handles, and box lids; be able to use six-axis force sensors and joint torque feedback to determine contact status and adjust control parameters online; practical experience with soft grippers, tactile sensors, and force-controlled grasping is preferred.

Track C: Bimanual Coordination and Complex Manipulation

• Understand the basic logic of coordinate transformation, synchronized trajectory, and collision constraints for dual robotic arms / humanoid dual-hand coordination; able to assist in implementing collaborative actions such as one hand fixed while the other operates, and dual-hand carrying and lifting; understand issues such as torque distribution in dual-hand operations, relative pose maintenance, and self-collision avoidance; able to encapsulate standardized dual-hand actions into reusable operational skills; experience in debugging dual-arm, humanoid, or mobile operational robots is preferred.

Track D: End-Effector and Task-Skill Library

• Capable of completing kinematic modeling and control interface adaptation for grippers, dexterous hands, suction cups, and long-handled operating tools; encapsulating actions such as grasping, pushing and pulling, pressing, and upright adjustment into standardized, reusable operational skills; iteratively optimizing end-effector control parameters based on task failure samples; collaborating with perception, motion control, and planning modules to achieve a complete closed loop from recognition to grasp execution; experience with operation skill libraries and demonstration data collection is preferred.

Preferred Qualifications

• Practical experience with projects related to dexterous hands, multi-fingered hands, grippers, and tactile sensors;
• Debugged humanoid, dual-arm, mobile robotic arms, and service robot physical operation tasks;
• Familiar with mainstream dexterous hands and gripper hardware such as Allegro Hand, Robotiq, and Franka Hand;
• Able to use simulation/kinematics tools such as MoveIt, Pinocchio, MuJoCo, Isaac Sim, and Gazebo;
• Involved in development related to visual servoing, haptic feedback, force-controlled grasping, and slip detection;
• Has completed physical robot tasks such as pushing doors, pressing buttons, operating tools, and carrying with both hands;
• There are practices related to supervised learning, imitation learning, reinforcement learning, and operational data set collection.
• Understand EtherCAT/CAN real-time bus, high-frequency closed-loop control, and lower-level motor drive debugging;
• Core member of RoboMaster and robotics competitions, participated in complete hardware and software robot projects;
• GitHub has open-source projects related to robotics, with clear code standards;
• Able to use AI tools such as Claude Code and Codex to assist in code and algorithm development.