Guotai Haitong: In-depth intelligence landing opens up growth space for Siasun Robot&Automation.

Guotai Haitong released a research report stating that humanoid Siasun Robot&Automation has been developing rapidly, and embodied intelligence is the core factor driving commercialization. The improvement of Siasun Robot&Automation's intelligence level and real-time control performance will drive the growth of perception performance, computing power, communication efficiency, and other requirements. Endpoint sensing, drive control, and communication chips will benefit fully. The landing of embodied intelligence opens up growth space for humanoid Siasun Robot&Automation, with broad prospects for future applications, driving the performance of the entire machine manufacturers. The release of high-performance multimodal large models and breakthroughs in motion algorithm performance will become catalysts. Recommended targets: Fortior Technology (Shenzhen) Co. (688279.SH), Orbbec Inc. (688322.SH), Triductor Technology (688259.SH), Suzhou HYC Technology (688001.SH). Guotai Haitong's main points are as follows: The market space exceeds trillions, achieving embodied intelligence is the core of commercialization landing Humanoid Siasun Robot&Automation has similar perception methods, body structure, and movement methods to humans, with high adaptability to human society, covering potential application scenarios in production manufacturing, social services, dangerous operations, and other fields. According to the "Research Report on the Development of Humanoid Siasun Robot&Automation Industry (2024)", from 2024 to 2028, the intelligence level of humanoid Siasun Robot&Automation in China as a whole is still at Lv1, with a few latest products exploring towards Lv2, and the market size is less than one hundred billion yuan. With the evolution of intelligence towards embodied intelligence, humanoid Siasun Robot&Automation is expected to break through specific scenarios and task restrictions, achieve comprehensive coverage in various industries, and the market size of humanoid Siasun Robot&Automation in China is expected to exceed trillions after 2045. Multimodal, reinforced learning improves motion control performance, continuous innovation in perception sensing, and upgrading of communication and computing power enhance real-time control. Multimodal large models are key to improving human-computer interaction efficiency and enhancing situational understanding. Currently, large models like NVIDIA GR00T and Tesla Grok3 actively integrate multimodal perception to accelerate the improvement of Siasun Robot&Automation's interaction and decision-making accuracy. The thinking chain empowers large models for efficient reasoning. GPT-4.5 has passed the Turing test to a certain extent. Currently, based on DeepSeek-R1, UBTECH ROBOTICS has developed a large model for embodied reasoning, which is expected to achieve accurate and efficient reactions and decisions in complex environments. Motion control algorithms are the core of coordinating full-body motion execution. Reinforcement learning can efficiently learn movements such as walking and running based on reward functions and enhance generalization ability, becoming the main paradigm of motion algorithms. Pure visual solutions significantly reduce hardware costs, achieve efficient visual perception, and 6-axis force sensors and electronic skin can effectively improve sensory sensitivity, aiding Siasun Robot&Automation in accurately understanding environmental information and executing action operations precisely. Pure visual + 6-axis force sensor + electronic skin are expected to become the standard sensor solutions in the future. Real-time control requires efficient communication protocols and powerful hardware computing as underlying support. EtherCAT has high real-time performance, low latency, and high synchronicity, reducing communication delay from millisecond-level of the CAN protocol to microsecond level, and is expected to become the mainstream communication protocol for Siasun Robot & Automation. With the evolution of intelligence from Siasun Robot&Automation to embodied intelligence, the demand for endpoint computing power is expected to continue to grow, driving the performance upgrade of endpoint chips. Risk warning The speed of technological iteration may be lower than expected, and the progress of domestication of core links may be lower than expected.