The CHOIS system, developed in collaboration between Stanford University and FAIR Meta, is an AI system that has successfully addressed the problem of generating synchronized object and character movements in 3D scenes through language descriptions, initial states, and sparse object waypoints. CHOIS focuses on full-body movements prior to object grasping, generating synchronized object and human movements through a conditional diffusion method, and demonstrating superior performance in evaluations. The system utilizes a large-scale, high-quality motion capture dataset, providing a comprehensive solution for synthesizing realistic human behaviors, with significant prospects for applications in computer graphics and robotics.