Robotics Software Engineer, Manipulation

Responsibilities

• Design and implement advanced, dexterous manipulation algorithms for humanoid robots with high DOF and multi-fingered hands.
• Develop and optimize the manipulation pipeline, encompassing perception, grasping, trajectory optimization, motion planning, and control.
• Tackle complex tasks such as precise grasping, dual-arm manipulation, and in-hand object manipulation.
• Implement robust control strategies for precise and dexterous manipulation, including force/motion control and visual/tactile servoing.
• Integrate the manipulation pipeline with the humanoid robot's whole-body controller and contribute to loco-manipulation development.
• Develop digital-twins of the manipulation scenes for algorithm development.
• Utilize simulators such as MuJoCo and Isaac Sim.
• Conduct rigorous testing in both simulated and real-world environments.
• Work closely with the machine learning team to help train and deploy models for advanced manipulation tasks.
• Stay informed about the latest research and advancements in manipulation, leveraging both traditional methods and deep learning-based approaches.

Requirements

• BS, MS, or PhD in Robotics, Computer Science, or a related field.
• 7+ years of experience (5+ with advanced degree) in robotics with a strong emphasis on control and manipulation of robotics/humanoid arms.
• Strong background in motion planning, control theory, and optimization.
• Demonstrated expertise in implementing advanced manipulation algorithms for tasks such as pick-and-place, bin picking, door opening, etc.
• Proficiency in C++, Python, and relevant robotics libraries (e.g., MoveIt!, Drake, OpenRAVE).
• Experience with trajectory optimization libraries (e.g., IPOPT, SNOPT, qpOASES).
• Proficiency with modern deep learning techniques for manipulation.
• Experience with computer vision algorithms, sensors, point clouds, segmentation and object detection.
• Familiar with ROS, LCM, or other middleware for robotic systems.

Preferred Qualifications

• Experience in developing algorithms for contact modeling and force/torque estimation with a focus on tactile sensing and feedback for dexterous tasks.
• Experience with machine learning for dexterous manipulation, including learning-based grasp strategies, behavior cloning and reinforcement learning.
• Publications in top-tier robotics conferences focusing on dexterous manipulation.
• Experience in a humanoid robot startup environment.