[Keynote] Physical AI for Dark Factories: Sim-to-Real Orchestration of Heterogeneous Robot Fleets Using Digital Twin Intelligence
10:05 am - 10:40 am
This presentation introduces Physical AI for dark factories, enabling autonomous orchestration of heterogeneous robot fleets, including AMRs, OHTs, 3D shuttles, and humanoid robots. Using a sim-to-real learning framework based on high-fidelity digital twins, robot intelligence is trained in virtual environments and seamlessly transferred to real manufacturing systems, significantly reducing deployment risk and commissioning time. Industrial case studies demonstrate large-scale fleet coordination, real-time optimization, and substantial improvements in throughput, congestion mitigation, and operational stability. The proposed framework establishes a practical pathway toward fully autonomous and scalable manufacturing systems.
