Our simulation support simulation of robots and human avatars. For avatars, we use 74 avatar skins representing normal individuals sourced from Mixamo and Academic, Industry, Sports, Politics, Entertainment, and Historic Celebrities. Additionally, it features 2,299 unique motions and 13 vehicles. We present 24 diverse community scenes, though infinite scenes could be generated. Each scene is automatically annotated with functional areas, including accommodation, entertainment venues, restaurants, offices, and stores, which also come with generated indoor scenes.
Below we present some demo videos of humanoid agents and robots walking and interacting in different geolocations around the world generated by our pipeline.
Agents are provided the interface to communicate and interact with each other and scene with a physics engine running behind the curtain. Each agent has its own navigation, scheduling, and memory system and has generated an identity profile to support social interaction at scale.
Manhatton,
New York City, New York, United States
Dever Probate Court Clerk,
Denver, Colorado, United States
The Monroe Street Midway
Detroit, Michigan, United States
Palace of Westminster,
London, United Kingdom
Palacio de Biblioteca y Museos Nacionales,
Madrid, Spain
Indipendenza Ridolfi,
Florance, Italy
We demonstrate the 3D model generated by our pipeline in cities all around the world. Our pipeline is able to create virtual scene at scale with photo-realistic view and simplistic geometry.
Amsterdam, Netherlands
New York, United States
Bratislava, Slovakia
El Paso, United States
Calgary, Canada
Hamburg, Germany
Columbus, United States
Florance, Italy
London, United Kingdom
Detroit, United States
Denver, United States
Charlotte, United States
More aerial view examples of outdoor scene generation results
Store
Dormitory
Restaurant
Office