Video2Act: Novel Robotic Action Generation via Video Diffusion

This ArXiv paper introduces Video2Act, a new approach to robotic action generation using video diffusion models. The dual-system design for spatio-motional modeling shows promise for improved robot control in complex tasks.

• •Proposes a novel approach for robotic action generation.
• •Utilizes a dual-system architecture for improved modeling.
• •Employs video diffusion models for generating actions.