First batch of national standards for humanoid robots released; Dobot participated in drafting two standards on motion control and task operation

Humanoid robots can walk, run, and pick things up, but to enter real-world scenarios such as factories and shopping malls, they still need to answer more specific questions: Are their movements accurate enough? Can they adjust in time when they encounter changes? When picking up different objects, can they grasp them steadily and place them accurately? How to evaluate these capabilities requires a set of standards recognized across the industry. Recently, the State Administration for Market Regulation (Standardization Administration) released the first batch of six national standards for humanoid robots, which will take effect on April 1, 2027. The standards cover general requirements, environmental perception, decision-making and planning, motion control, task operation, and simulation testing, providing a common basis for robot R&D, capability evaluation, and test verification. As a participating drafting unit, Dobot deeply participated in the drafting of two standards: "Technical Requirements for Humanoid Robots Part 4: Motion Control" (GB/T 48317.4—2026) and "Technical Requirements for Humanoid Robots Part 5: Task Operation" (GB/T 48317.5—2026). These two standards focus precisely on how humanoid robots coordinate their bodies and how they complete work. Participating in the drafting of two core standards, focusing on the execution and operation capabilities of humanoid robots. After a robot understands instructions, it still needs to make its body execute them accurately. For example, when reaching out to pick up an object, the arm moves forward, and the torso and legs must also coordinate to maintain balance; when turning and walking, multiple joints need to respond in coordination to make movements stable and coherent. "Motion Control" focuses on these bodily movement capabilities. "Task Operation" further focuses on whether tasks can be completed. For the same grasping action, picking up a metal part and picking up a soft object require different postures and forces; placing an object into a material box and inserting a part into a designated position also differ in precision requirements. Simply put, motion control enables robots to act in a coordinated and accurate manner, while task operation enables these actions to produce actual results. With common standards, R&D companies can more clearly determine technical requirements, testing institutions can evaluate accordingly, and customers can more accurately understand what a robot is suitable for and how its capabilities can be verified. From motion control to dexterous operation, supporting standard co-construction with a decade of technological accumulation. Dobot's exploration of motion and operation capabilities began with ten years of deep cultivation in the robotics field. With more than 100,000 robots deployed globally, serving more than 100 Fortune Global 500 companies, and covering 15 major industries and more than 200 segmented scenarios, Dobot has continuously refined motion control, force control, trajectory planning, and safe interaction capabilities in real industrial environments, accumulating a solid technical and engineering foundation for participating in the drafting of the two standards. These capabilities come from the specific requirements of production lines for operation quality and efficiency: maintaining accurate trajectories during high-speed motion, suppressing vibration during continuous start-stop, achieving precise synchronization during multi-joint operation, and coordinating position and force when contacting workpieces. Around these needs, Dobot has continuously developed high-performance motion control algorithms and verified and optimized them in different processes and working conditions. For humanoid robots, motion control also needs to further address full-body multi-degree-of-freedom coordination and dynamic balance under changes in posture, load, and external disturbances. Based on existing control algorithms, joint drive, and engineering experience, Dobot has self-developed a high-precision full-body force control system, achieving millisecond-level dynamic balance and agile movement, extending control capabilities from the precise execution of robotic arms to full-body coordination, and providing support for stable movement and operation in complex environments. After making the body coordinate, it is also necessary to make both hands do the work well. When facing fragile, soft, or irregular objects, robots need to determine where to grasp and how much force to use, and adjust their actions based on contact conditions. Dobot has self-developed the Kongyi embodied large model, which connects natural language instruction understanding, autonomous planning, and task execution; at the same time, it has self-developed the Neural Drive Dexterous System (NDS), combining millimeter-level operation precision with adaptive force control to support multiple types of tasks from precision assembly to gentle grasping. From underlying control to full-body coordination, and then to fine operation, complete R&D and application experience enables Dobot to analyze problems by combining specific technical links with overall machine performance, providing engineering references for capability descriptions, evaluation conditions, and verification requirements in standard drafting. Turning technological advantages into industrial advantages and driving China's embodied intelligence forward. Global competition in embodied intelligence is testing the synergy among technological innovation, manufacturing capability, and scenario application. China has a complete manufacturing foundation, rich application demands, and active innovation forces. To transform these conditions into sustainable industrial advantages, enterprises need to continuously open up the path from independent R&D and product supply to commercial delivery, and participate in the construction of common technical rules. Dobot is participating in this process with a multi-form product layout and commercialization practices. It has built a product matrix covering collaborative robots, bipedal humanoid robots, wheeled humanoid robots, and multi-legged robot dogs, relying on the self-developed Kongyi embodied large model to advance the commercialization of related products, and has entered the mass production stage of humanoid robots. As of now, Dobot has cumulatively expanded to more than 230 embodied intelligence customers, covering fields such as industrial manufacturing, education and research, and commercial retail. Related projects are at different stages, including order signing, POC verification, small-batch trial production, and large-scale batch delivery. These practices allow products to be tested against real demands and also accumulate experience for deployment, operation, and subsequent application replication. Put products into real scenarios and bring experience into standard construction. Dobot's participation in the drafting of two national standards is precisely about bringing R&D and application accumulation into industrial collaboration, so that problems discovered in Chinese scenarios and insights formed in Chinese engineering practice can provide support for improving industry technical requirements. Looking ahead, Dobot will continue to advance independent innovation, commercial delivery, and standard co-construction, working with industry partners to transform the engineering accumulation and scenario advantages of Chinese manufacturing into industrial competitiveness in the era of embodied intelligence. Original title: The first national standards for humanoid robots, with Dobot writing them