PhysDiff: Physics-Guided Human Motion Diffusion Model
This addresses the limitation of existing motion diffusion models for realistic human motion generation, enabling better real-world applications in fields like animation and robotics, though it is an incremental improvement by adding physics guidance to an existing paradigm.
The paper tackles the problem of generating physically-implausible human motions with artifacts like floating and foot sliding in diffusion models, and presents PhysDiff, which incorporates physical constraints to achieve state-of-the-art motion quality and improves physical plausibility by over 78% across datasets.
Denoising diffusion models hold great promise for generating diverse and realistic human motions. However, existing motion diffusion models largely disregard the laws of physics in the diffusion process and often generate physically-implausible motions with pronounced artifacts such as floating, foot sliding, and ground penetration. This seriously impacts the quality of generated motions and limits their real-world application. To address this issue, we present a novel physics-guided motion diffusion model (PhysDiff), which incorporates physical constraints into the diffusion process. Specifically, we propose a physics-based motion projection module that uses motion imitation in a physics simulator to project the denoised motion of a diffusion step to a physically-plausible motion. The projected motion is further used in the next diffusion step to guide the denoising diffusion process. Intuitively, the use of physics in our model iteratively pulls the motion toward a physically-plausible space, which cannot be achieved by simple post-processing. Experiments on large-scale human motion datasets show that our approach achieves state-of-the-art motion quality and improves physical plausibility drastically (>78% for all datasets).