ARDY is a streaming human-motion generation framework designed for interactive applications such as animation, simulation, and humanoid robotics. The method combines explicit root features with a latent body representation, then uses a two-stage autoregressive Transformer denoiser with variable history context. It supports online text prompts and flexible, long-horizon kinematic constraints, including keyframe poses and path-following controls. The authors evaluate ARDY on HumanML3D and the high-fidelity Bones Rigplay dataset, and present an interactive demo covering dynamic text control, locomotion, and keyboard or mouse input. The supplied abstract does not report detailed numerical results.
No heat snapshots are available in the last 24 hours.