The Challenge Pilot Addresses

  • High variability across brands, models, sizes, and damage levels.
  • Worker safety risks: repetitive strain, sharp edges, heavy parts, electrical hazards.
  • Inconsistent throughput and limited process efficiency.
  • Poor data capture for logging and KPI tracking.

A human-in-the-loop disassembly cell that fuses generative pose estimation, three control modes (Autonomy / Shared Control / Teleoperation with safety assist), and adaptive safety compliant with ISO/TS 15066. The cell executes consistent, traceable disassembly tasks, running AI locally on-site for fast response times and secure, sovereign data handling.

Objectives

The pilot develops and tests an EARTH cell that fuses generative pose estimation, 3-mode
shared control, and ISO/TS 15066 adaptive safety.

EARTH_JARVIS_one_pager_Sprint3Download

Key Technologies or Methods Being Developed or Tested

Verifiable material-flow data enabling safety, efficiency, and sustainability KPIs.

  • –50% worker hazard exposure.
  • Reduction of reconfiguration & programming time by 30%.
  • Operator SUS >60%.
  • 2 JARVIS tools at TRL-7: Vision (object recognition & pose estimation) and Teleoperation.
  • Plug-and-play integration via ROS 2 + Docker + OPC UA.

EARTH Pilot Project – Sprint #3