Recently, CVPR 2026, one of the premier international conferences dedicated to computer vision, unveiled its standout research accomplishments. A collaborative research team from Peking University, Shangwei Qiyuan Institute, and JD.com has introduced the RealAppliance dataset, along with its corresponding evaluation benchmark, RealAppliance-Bench. This groundbreaking work centers on the intricate challenge of devising intelligent operation plans for household appliances based on instruction manuals. By leveraging a high-fidelity simulation system, the research effectively tackles the hurdles in practical evaluation for home service robots, thereby expediting the deployment of embodied AI applications in real-world home environments.
