Embodied intelligence products come in a variety of types. To meet the simulation needs of different types and application scenarios, emerging embodied intelligence simulation platforms are gradually entering the market.

The embodied intelligence simulation platform is a simulation platform used to study embodied intelligence. It provides a safe and controllable experimental environment to simulate various agent behavior patterns and complex application scenarios, including family life, industrial production, outdoor operations, business transactions, social interactions, and other scenarios.

Embodied intelligence is a cutting-edge technology field that integrates artificial intelligence, robotics, computer vision, and other disciplines. It has been included in the "Alternative List of New Productivity Technologies for New-generation Information Engineering Technology" (2024) released by the Chinese Academy of Engineering. In terms of commercialization, several companies such as Zhiyuan Robotics, MicroEye Intelligence, Unitree Robotics, and UBTECH have already established a presence in the embodied intelligence market, with an increasingly diverse range of related products.

According to the "2024-2028 China Embodied Intelligence Simulation Platform Industry Market Supply and Demand Status and Development Trend Forecast Report" released by the Xin Sijie Industry Research Center, the embodied intelligence simulation platform has high application value in the research and development, scalability training, and safety assessment of embodied intelligence products such as embodied intelligence robots, embodied intelligence autonomous vehicles, embodied intelligence cameras, and embodied intelligence brains. As the attention on embodied intelligence increases, the application space for embodied intelligence simulation platforms will continue to expand.

Embodied intelligence simulation platforms are divided into two main categories: general-purpose platforms based on underlying simulations and simulation platforms based on real-world scenarios. General-purpose platforms based on underlying simulations include NVIDIA Isaac Sim, AirSim, Pybullet, Unity, etc. Simulation platforms based on real-world scenarios include iGibson 1.0, TDW, SAPIEN, Habitat, etc.

Embodied intelligence products come in a variety of types. To meet the simulation needs of different types and application scenarios, emerging embodied intelligence simulation platforms are gradually entering the market. For example, in 2024, the Shanghai Artificial Intelligence Laboratory (Shanghai AI Lab) released the Puyuan Taoyuan Simulation Platform GRScenes, the first city-level embodied intelligence simulation platform. The Tsinghua University Urban Science and Computing Research Center released the EmbodiedCity, an urban embodied intelligence simulation environment platform based on Unreal Engine 5, which offers both online access and offline operation modes.

In addition, relevant research and development units of embodied intelligence simulation platforms in China include the Beijing Embodied Intelligence Robotics Innovation Center (which has been upgraded to the National-Local Collaborative Embodied Intelligence Robotics Innovation Center), the Guangdong Embodied Intelligence Robotics Innovation Center, and the Shenzhen Institute of Artificial Intelligence and Robotics (AIRS).

Industry analysts at Xin Sijie stated that embodied intelligence is the next blue ocean in the field of artificial intelligence, with great application potential in industrial manufacturing, logistics and transportation, household services, medical healthcare, and other fields. However, the engineering implementation of embodied intelligence is complex, with issues such as slow training/testing speeds, high costs, and difficult-to-control experimental environments in the real world. The embodied intelligence simulation platform can provide a safe and controllable experimental environment and has high application value in the field of embodied intelligence research and development. In the future, embodied intelligence simulation platforms will continue to be optimized and upgraded to meet diverse simulation needs.