Locomotion & Whole-Body Control Engineer (human)

NEURA Robotics; Riederich; Vor Ort; EUR 90.000 - 130.000; Vollzeit

Zusammenfassung

NEURA Robotics is expanding its legged-robotics capability on the humanoid (4NE1) and quadruped platforms. This role spans core layers of the legged control stack, including trajectory optimization and MPC for kino-dynamic motion generation, and QP-based whole-body control at 1 kHz on real hardware.

You will collaborate with state estimation, simulation, low-level control, and hardware teams to ensure robust, predictable locomotion and whole-body behavior across tasks and platforms.

Qualifikationen

• MSc or PhD in robotics, controls, mechanical or electrical engineering, computer science, or a related field.
• 4+ years of hands-on experience developing trajectory optimization, MPC for locomotion, and/or whole-body control on physical robots.
• Strong foundation in floating-base articulated rigid-body dynamics and contact modelling.
• Strong working knowledge of reduced-order locomotion models (LIPM, SRBD, centroidal dynamics, or equivalents) and their use inside MPC.
• Strong foundation in optimal control, constrained numerical optimization, and model-predictive control for legged robots.

Aufgaben

• Whole-body motion generation and control for floating-base legged platforms—locomotion, balance, contact transitions, and loco-manipulation.
• Trajectory optimization and MPC pipelines over robot state, contact schedules, ground reaction forces, centroidal momentum, and joint trajectories.
• QP-based task-space inverse dynamics for executing instantaneous whole-body control from MPC/trajectory references at 1 kHz on real hardware.
• Whole-body modelling for the platform: floating-base rigid-body dynamics, URDF/MJCF, Jacobians, and dynamics computations.
• Constraint formulation across MPC and QP layers with contact, friction, torque, joint, and stability constraints.

Kenntnisse

C++ for real-time robotics

Python scripting

Floating-base dynamics

Contact modelling

MPC/trajectory optimization

Whole-body control / TSID

Collaboration across disciplines

Ausbildung

MSc or PhD in robotics/controls/mechanical or related

Tools

Pinocchio

MuJoCo

Crocoddyl

IPOPT

TSID

OCS2

Jobbeschreibung

Your mission & challenges

We are growing our legged-robotics capability on NEURA's humanoid (4NE1) and quadruped platforms. This role spans both core layers of the legged control stack: trajectory optimization and MPC for kino‑dynamic motion generation, and QP‑based instantaneous whole‑body control for executing those motions on real hardware at 1 kHz.

The work is focused on contact‑rich dynamics, real‑time optimization, and reliable execution on physical robots. You will collaborate closely with state estimation, simulation, low‑level control, and hardware stakeholders, and with the application teams whose tasks ultimately depend on robust, predictable locomotion and whole‑body behaviour.

• Whole‑body motion generation and control for floating‑base legged platforms — locomotion, balance, contact transitions, and loco‑manipulation (walking while manipulating).
• Trajectory optimization and model‑predictive control pipelines over robot state, contact schedules, ground reaction forces, centroidal momentum, and joint trajectories — using reduced‑order locomotion models such as LIPM, SRBD, and centroidal dynamics.
• QP‑based task‑space inverse dynamics for executing instantaneous whole‑body control from MPC and trajectory‑optimization references at 1 kHz on the real robot.
• Whole‑body modelling for the platform: floating‑base rigid‑body dynamics from URDF / MJCF, joint configuration, FK / IK, Jacobians, and mass / Coriolis / gravity computation.
• Constraint formulation across the MPC and QP layers — contact, friction, torque, joint, kinematic, and stability constraints — with task‑hierarchy design appropriate to the platform.
• Solver performance work across both layers: warm‑starting, numerical conditioning, constraint handling, and real‑time reliability at 500 Hz – 1 kHz.
• Deployment, tuning, and debugging of MPC, trajectory optimization, IK, and inverse dynamics pipelines on physical robots — including platform‑specific contact‑model calibration and validation against real robot data.
• High‑performance C++ for real‑time execution; Python tooling for analysis, prototyping, and debugging.

What We Can Look Forward To

• MSc or PhD in robotics, controls, mechanical or electrical engineering, computer science, or a related field.
• 4+ years of hands‑on experience developing trajectory optimization, MPC for locomotion, and / or whole‑body control on physical robots.
• Strong foundation in floating‑base articulated rigid‑body dynamics and contact modelling.
• Strong working knowledge of reduced‑order locomotion models (LIPM, SRBD, centroidal dynamics, or equivalents) and their use inside MPC.
• Strong foundation in optimal control, constrained numerical optimization, and model‑predictive control for legged robots.
• Hands‑on experience with whole‑body QP / TSID frameworks on real robot data — including QP / DDP solver internals.
• Hands‑on experience deploying real‑time control / MPC / WBC pipelines at 500 Hz – 1 kHz on hardware.
• Strong C++ for real‑time robotics software; Python for analysis, tooling, prototyping, and debugging.
• Practical understanding of how contact dynamics, actuator limits, latency, state‑estimation error, solver failure modes, and model mismatch behave on real hardware.
• A collaborative working style: shared design, constructive code review, proactive communication, and reliable coordination across control, estimation, simulation, low‑level control, and hardware disciplines. Strong teamwork is essential for this role.

Nice to Have

• Hands‑on experience on humanoids, quadrupeds, or other high‑DOF legged robots.
• Familiarity with Pinocchio, MuJoCo, Crocoddyl, IPOPT, TSID, OCS2, or similar open‑source tools.
• Hierarchical QP, weighted QP, task prioritization, contact force optimization, or operational‑space control.
• Contact planning, gait optimization, balance recovery; CPG‑based or hybrid CPG / MPC controllers.
• Multi‑contact WBC: foot contact, bimanual grasping, or base‑arm coordination.
• Contact‑consistent dynamics and impact‑aware control transitions.
• Experience with torque‑controlled robots and high‑bandwidth electric actuation.
• Publications at RSS, ICRA, IROS, or CoRL in legged locomotion or whole‑body control.