By EVCar 
The humanoid robot series announced by the research institute this time includes 7 products, which form different robot positioning according to the differences in application scenarios, covering bipedal, wheeled, quadruped, tracked and other forms.
Based on the adaptive characteristics of the scene, the team from Youai Zhihe Xi’an Jiaotong University embodied intelligent robot research institute has constructed a “one brain polymorphism” embodied intelligent large model, using a hybrid architecture of a multimodal universal base large model and a “one brain polymorphism” end side embodied model, and completed preliminary verification of the scene application.
Under the “One Brain Multi state” End to Side Embodied Model Architecture, the team has integrated key technologies such as reinforcement learning, imitation learning, and deep learning to create a embodied “brain” MAIC (Mobile AI Complexity) for multimodal interaction, visual semantic perception, and behavioral reasoning decision-making. This has resulted in end-to-end control of the embodied “limbs” for body perception, whole body joint control, and agile operation, as well as multi-modal sensing fusion based polymorphic “movement” capabilities for different scenarios.
Based on the technology and market accumulation of Youai Zhihe in the industrial field, the research institute has released a humanoid robot series that aims to generalize industrial scenarios and accelerate the arrival of embodied intelligence for large-scale industrial applications.