HUMANOID CONTROLS

Internship ProjectTeam
Arm integration, Ethernet teleoperation, and headset-driven neck controls.
Date 08 September 2026
Role Controls Engineer — Mundane Systems Inc.
Contribution Software development
Skills C++, Python, ROS 2, Pinocchio, URDF, CAN, Gravity compensation, PD control, Hardware integration

At Mundane Systems, I helped bring new robot arms into the existing leader–follower system and tuned the headset-driven neck.

Arm bring-up

  • Integrated URDFs and calibrated joint mappings, encoder directions, offsets, and torque limits for the new seven-DOF HighTorque arms.
  • Tuned Pinocchio gravity compensation through single-joint and full-arm float tests, supporting 1 kHz local control over CAN FD.
  • Characterized Coulomb friction and developed compensation profiles; some joints still needed validation.

Ethernet teleoperation

  • Configured cross-machine ROS 2 communication, clock synchronization, and launch and diagnostic tools.
  • Restored leader–follower tracking by fixing false dropout detection and holding the latest ~100 Hz reference between updates to the 1 kHz local controller.
  • Added command-age checks to reject stale references and a controlled transition to a held pose on link loss.

Headset-driven neck

  • Converted headset orientation into calibrated joint targets for the three-DOF Damiao neck.
  • Tuned PD control, orientation smoothing, and velocity/acceleration limits for smoother motion.
  • Fitted sinusoidal gravity feedforward and built calibration and motion-logging tools.

Watch the public customer demo at NECA 2026 on LinkedIn →