2026 World AI Device Expo | Highlights of VisIncept New Product Launch & Founder's Keynote Speech
The 2026 World AI Device Expo, alongside the 7th Shenzhen Global AI Expo, grandly kicked off in Shenzhen from May 14 to 16, 2026. The event featured numerous highlights. Zhang Lei, Founder & CEO of Visincept and Chair Scientist at IDEA Research Institute, delivered a keynote speech on world models as an invited guest. Meanwhile, Visincept joined hands with Baidu Intelligent Cloud to officially launch EgoTwin, the world's cutting-edge high-quality Ego 3D hand alignment engine. This launch stood out as a landmark achievement in the AI sector at this year's expo.
Video 1: Highlights of Visincept at the Expo
Highlight 1: Official Launch of EgoTwin 3D Hand Alignment Engine to Empower AI in the Physical World
The newly unveiled EgoTwin engine addresses a key bottleneck in the AGI industry: the shortage of high-quality real-world 3D alignment data for first-person (egocentric) hand scenarios. Powered by first-person video data, it achieves precise 3D hand keypoint localization and hand mesh reconstruction, accurately restoring details of real hand interactions and spatial logic.
Leveraging its proprietary core technology for high-precision 2D-to-3D alignment, EgoTwin efficiently converts egocentric data to adapt to robot state spaces. It drastically improves data augmentation efficiency compared with traditional approaches such as physical robot teleoperation and UMI.
Figure 1: Joint launch of EgoTwin Data Engine by Visincept and Baidu Intelligent Cloud
This collaboration between Visincept and Baidu Intelligent Cloud aims to further build industrial ecosystems and iterate cutting-edge technologies, diversify intelligent application solutions across multiple fields, accelerate the large-scale deployment of general embodied intelligence, and drive sound and high-quality development of physical AI.
Highlight 2: Founder Shares In-depth Insights on World Models, Outlining Corporate Vision and Tech Roadmap
At the industry forum On the Eve of AGI: The Awakening of Large Language Models, Zhang Lei delivered a speech titled From Perception to Prediction: The Dawn of World Models and Super Physical Intelligence. He thoroughly analyzed the paradigm shift of AI expanding from the digital realm to the physical world, as well as the demands, evolution and core essence of world models.
Figure 2: Zhang Lei, Founder & CEO of Visincept, delivers a keynote on world models
During the speech, Zhang Lei emphasized the core value of object-centric and action-aligned latent world models. Boasting solid technical expertise and top-tier research capabilities in 3D motion interaction and 3D object understanding, Visincept will partner with leading AI enterprises to push forward the commercialization of underlying infrastructure for world models, helping AI truly bridge the gap between the digital and physical worlds.