Views: 0 Author: Site Editor Publish Time: 2025-04-30 Origin: Site
On April 25, the China Humanoid Robot Ecological Conference opened in Shanghai. The event is themed "Human Machine Integration, Intelligent Manufacturing for the Future". Liu Yaodong, Chairman and CEO of Virtual Dynamic Point Company, was invited to attend and give a keynote speech. At the meeting, he had in-depth discussions with guests to discuss industry development.
The event attracted numerous leading companies in the industry chain, including Yushu Technology, Leju Robotics, Agile Intelligence, Time Robot, and Xingdong Era, to attend. It gathered top industry forces, authoritative experts, leading enterprises, investment institutions, and industry media to jointly explore the direction of humanoid robots and embodied intelligence.
In the keynote speech section of the conference, Liu Yaodong elaborated on the core capabilities of virtual motion points in the field of embodied intelligence, with the theme of "Space Computing Empowering Robot Training", focusing on the development direction of embodied intelligence technology that the industry is concerned about.
When AI breaks through the boundaries of the virtual world, embodied intelligence finally breaks free from conceptual constraints, with the key being "data". The 2025 Research Report on Humanoid Robots and Embodied Intelligence Industry predicts that high-quality data integration will become an accelerator for embodied intelligence. During the event, Liu Yaodong solved the problem of embodied intelligence "data" for numerous industry partners.
He stated that the virtual motion point database has a large amount of high-quality high-precision humanoid and robot data, and can provide customized data collection services, especially adept at collecting key data in real scenarios. And it can provide various services including end of pipe data collection, simulation training, remote operation training, etc.
Liu Yaodong also mentioned that virtual motion points continue to stand on a solid foundation of spatial computing, upgrading the world's leading OptiTrack optical algorithm products, LydCap label free vision algorithm, LYDIA action big model products, humanoid action database, etc., and launching optical and label free fusion algorithms globally, which can provide high-precision and convenient action data collection and algorithm support for robot training at the same time.
Especially the evolution of the virtual motion point LYDIA action model provides a new solution for the continuous upgrading of embodied intelligent action databases. Liu Yaodong introduced that with the integration of 20 billion parameters and DeepSeek language model, LYDIA not only further improves its understanding of action intentions in textual descriptions, but also has the ability to "generate actions and interact with the environment". It can autonomously plan the path of virtual characters in complex scenes in the model and interact with specified targets. LYDIA opens up new scenarios for obtaining action data of humanoid robots in a convenient and efficient way.
At the meeting, Liu Yaodong also presented to the guests the achievements and cases accumulated by virtual motion points in the field of embodied intelligence and global partners. With the support of virtual motion point gait data and services, the partner robot teams completed the world's first half marathon robot race in second and third place respectively.
At the end of the event, Liu Yaodong stated that Virtual Point will stand on a solid foundation of spatial computing, continue to empower the embodied intelligence industry, help partners create better robots, and accelerate the commercialization of humanoid robots in specific work and life scenarios such as homes, enterprises, and industries. He looks forward to working together with more partners to innovate machines!
