Beijing: Concentrate on breakthroughs in key technologies such as the universal prototype of humanoid robots and the large model of general artificial intelligence

According to a report by the Financial Associated Press on June 28, the General Office of the Beijing Municipal People's Government recently issued the "Beijing Robot Industry Innovation and Development Action Plan (2023-2025)". Among them, it is proposed to step up the layout of humanoid robots. Benchmark the world's leading humanoid robot products, support enterprises and colleges and universities to carry out research and engineering of humanoid robot products and key components, accelerate the construction of Beijing Humanoid Robot Industry Innovation Center, and strive to create a national manufacturing innovation center. With the goal of small-batch production and application of humanoid robots, create basic conditions such as general intelligent underlying software and interfaces, and general hardware development supporting facilities, focus on making breakthroughs in key technologies such as general prototypes of humanoid robots and large models of general artificial intelligence, and vigorously promote open source control systems , open source chips, open source simulation software and other development and application.

In addition, strengthen the artificial intelligence large model support. Develop and continue to improve the general artificial intelligence model of robots, mine application scene resources, provide diversified scene data support for model pre-training, and improve the versatility and practicability of the model. Make breakthroughs in the key technology of large-scale multi-mode data fusion, and develop large-scale model systems for fusion processing of multi-mode sensor data such as images, text, voice, force, heat, electricity, and magnetism. According to the characteristics of various robotic technologies and application scenarios, develop efficient fine-tuning algorithms for large models, and promote the in-depth application of large models in the field of robotics. Build an open-source platform for model optimization algorithms, create a large-scale model ecosystem with extensive participation and interactive optimization across the industry, and promote iterative improvement of model performance.