AI for Industry Challenge by Intrinsic and Open Robotics (www.intrinsic.ai)

Intrinsic and Open Robotics have launched the AI for Industry Challenge — a multi-phase competition (Feb 11–July 2026) that tasks teams with automating dexterous cable assembly and wiring of server trays, a canonical hard problem in robotics due to deformable-object handling, connector variety, and high-cost failure modes. Registration closes April 17, 2026; teams (up to 10 members, one team per person) compete for $180,000 in prizes (top prize $100k). The challenge moves from simulation (qualification) to a Flowstate development phase and finally to real-world deployment on an Intrinsic-provided workcell, spotlighting solutions that can bridge sim-to-real gaps and scale to industrial settings. Participants receive a detailed toolkit: SDFormat scene descriptions, high-fidelity URDF/SDF robot and sensor assets, a baseline Gazebo environment, standardized ROS interfaces, a reference controller and hardware-abstraction layer, and a formal task spec. Qualified teams gain access to Intrinsic Flowstate and the Intrinsic Vision Model (IVM) — a transformer-based perception foundation — to integrate, debug, and validate end-to-end systems. Evaluation is automated (model validity on ROS topics, per-insertion task success, precision, safety penalties for collisions/forces, and cycle-time efficiency) with a final expert-judge review for innovation, technical soundness, and scalability. The challenge is a practical proving ground for ML-driven perception, control, and sim-to-real techniques in industrial manipulation.