Embodied AI refers to artificial intelligence integrated into physical systems, such as robots, enabling real-world interaction through sensors and actuators. Unlike traditional AI, which operates digitally, embodied AI learns from real-world experiences, adapting to dynamic environments and improving decision-making over time. This field combines machine learning, robotics, computer vision, and natural language processing to create intelligent agents capable of perception, reasoning, and physical interaction. Applications range from autonomous vehicles and service robots to healthcare assistants and industrial automation, revolutionizing various industries. However, challenges remain in developing robust learning algorithms, ensuring safety in unpredictable settings, and addressing ethical concerns related to autonomy and decision-making.
Video generation models can look physically convincing while getting the physics completely wrong. PhyCo, new research from our Media Analytics department, introduces continuous, controllable physical properties to video AI, allowing practitioners to specify friction, bounce, and force.
Modern video diffusion models excel at appearance synthesis but still struggle with physical consistency: objects drift, collisions lack realistic rebound, and material responses seldom match their underlying properties. We present PhyCo, a framework that introduces continuous, interpretable, and physically grounded control into video generation. Our approach integrates three key components: (i) a large-scale dataset of over 100K photorealistic simulation videos where friction, restitution, deformation, and force are systematically varied across diverse scenarios; (ii) physics-supervised fine-tuning of a pretrained diffusion model using a ControlNet conditioned on pixel-aligned physical property maps; and (iii) VLM-guided reward optimization, where a fine-tuned vision-language model evaluates generated videos with targeted physics queries and provides differentiable feedback. This combination enables a generative model to produce physically consistent and controllable outputs through variations in physical attributes–without any simulator or geometry reconstruction at inference. On the Physics-IQ benchmark, PhyCo significantly improves physical realism over strong baselines, and human studies confirm clearer and more faithful control over physical attributes. Our results demonstrate a scalable path toward physically consistent, controllable generative video models that generalize beyond synthetic training environments.