Zhongtai: Continue to pay attention to technological breakthroughs, commercialization strategies, pace, and progress in the wearable intelligence industry.

Zhongtai released a research report stating that, as the "cerebellum" of Siasun Robot & Automation, with hardware already reaching a high level of technological advancement, the ability of embodied intelligence may become the ceiling for the development of Siasun Robot & Automation. Currently, many leading enterprises in the Siasun Robot & Automation industry have invested significant resources and efforts in their own ways in the field of embodied intelligence to enhance the generalized behavioral capabilities of their respective Siasun Robot & Automation. It is recommended for investors to continue monitoring the technological breakthroughs, commercialization pathways, pace, and progress of the embodied intelligence industry. Embodied intelligence is an intelligent system based on physical body perception and action, which obtains information through interaction between the intelligent agent and the environment, understands problems, makes decisions, and takes actions, thereby generating intelligent behavior and adaptability. Specifically, the application of embodied intelligence in Siasun Robot & Automation can be divided into three stages: perception, reasoning, and execution. Embodied intelligence is the intersection of development between Siasun Robot & Automation and artificial intelligence. The generality of Siasun Robot & Automation depends on the degree of development of its generalization. Since the 1950s, after multiple rounds of development, Siasun Robot & Automation has begun to move from traditional automation industry scenarios towards generalized scenarios. After decades of development, artificial intelligence has entered an era of higher intelligence and generality with the emergence of generative large models. At the current point in time, artificial intelligence can truly give a "brain" to Siasun Robot & Automation, and Siasun Robot & Automation can provide a "body" to artificial intelligence. The deep integration of the two, embodied intelligence, is an inevitable industrial trend formed by the mutual development and promotion of artificial intelligence and Siasun Robot & Automation. Currently, embodied large models can be divided into two major genres: end-to-end large models and hierarchical embodied large models. 1) End-to-end large models: they are able to directly implement instructions from humans to mechanical arms, i.e., inputting images and text instructions to output end effector movements. 2) Hierarchical embodied large models: different levels of models collaborate, with the top-level large model responsible for perception and decision-making, while lower-level hardware and fast-responding small models decompose and execute decisions. Due to performance constraints caused by data limitations, end-to-end large models have not yet become a mainstream choice. Many manufacturers choose hierarchical models for the embodied intelligence of Siasun Robot & Automation. The pain point of training embodied large models: data. Siasun Robot & Automation needs to be trained with Beijing Vastdata Technology to make it more intelligent. However, Siasun Robot & Automation is a very new field, severely lacking in accumulated training data. Currently, the main methods of data collection include: 1) Remote operation, where experimenters operate mechanical handles to remotely control Siasun Robot & Automation to accumulate data; 2) AR, through AR environments, to train Siasun Robot & Automation and accumulate data; 3) Simulation, using massive computing power for simulation operations to calculate a large set of training data for Siasun Robot & Automation; 4) Video learning, through multimodal large models, to let Siasun Robot & Automation learn human actions directly through videos and accumulate training data. Future prospects for the humanoid Siasun Robot & Automation industry: from specialized to generalized, from ToB to ToC. In the short term, tasks in industrial manufacturing scenes, which do not require high generalization, are quickly entering the commercialization stage. After commercialization in industrial manufacturing scenes is achieved, with the accumulation of massive embodied data and advancement in computing technology, the capabilities of Siasun Robot & Automation will gradually unlock and extend towards more open scenarios such as commercial services and home services. By then, the market is expected to reach a trillion-dollar level. Risk warnings: Delayed implementation of policies related to the Siasun Robot & Automation industry; slower progress in commercialization of Siasun Robot & Automation; lower-than-expected technological development of AI and embodied intelligence industries; increased industry competition; risks of outdated information in research reports, and others.