Giving up her permanent overseas position, a female academician of the Chinese Academy of Sciences returned to China to lead a team to build a powerful national weapon

The humanoid robot market is very competitive, and many companies are trying to attract money and roll up their sleeves to get started.

While technology companies are still preparing ammunition, the national team, the Institute of Automation of the Chinese Academy of Sciences, has brought a heavy depth bomb, which may change the mode of humanoid robots from design to production, so as to quickly design and manufacture the most suitable humanoid robots according to changes in usage scenarios.

According to Science and Technology Daily, the team led by Academician Qiao Hong from the Institute of Automation of the Chinese Academy of Sciences has successfully developed a "big factory" for the design and assembly of humanoid robots, which can quickly design and build humanoid robot hardware and software systems. They have currently designed multiple humanoid robot prototypes and have initially achieved technical verification for different scenarios.

Give up permanent overseas positions and return to China to contribute to women's strength

This scientific research result comes from the team of Academician Qiao Hong from the Institute of Automation, Chinese Academy of Sciences. Academician Qiao Hong was elected as an academician of the Information Technology Science Department of the Chinese Academy of Sciences in 2021. In January this year, Academician Qiao Hong served as the director of the Expert Committee of the Beijing Humanoid Robot Innovation Center.

Academician Qiao Hong was born into a family of scholars. Her parents are professors at Xi'an Jiaotong University, which is also her alma mater. In 1989, Qiao Hong graduated from Xi'an Jiaotong University with a master's degree in computer science and technology. She then went to study in the UK and obtained a permanent teaching position and doctoral supervisor at the University of Manchester in 2002.

At that time, Qiao Hong's academic achievements were emerging overseas. She proposed the "environmental attraction domain" method in robot operation in the early stage, which was reported by international journals as "Qiao's concept". In her eyes, an important trend in the future development of robots is to become closer and closer to biological beings, no longer cold machines, but with a series of human-like characteristics including cognition, emotion, logic, and action.

In an interview, Academician Qiao Hong mentioned that he felt more at ease and satisfied when he made achievements in his own country. "Our country is in a critical historical period of development and growth. It is a Chinese's responsibility and honor to serve the motherland after learning."

Qiao Hong's research direction is the cross-integration of robotics and neuroscience. He leads his team to engage in long-term research on the "hand-eye-brain" of robots, namely the control of manipulators, robot vision, and brain-like artificial intelligence.

Academician Qiao Hong once said: "Helping to solve industry bottlenecks is the driving force that supports me to move forward." Faced with the wave of embodied intelligence, she and her team are bound to improve the comprehensive strength of my country's robots in this great revolution of artificial intelligence and make our robots at the forefront of the world.

With rapid design and construction in a “big factory”, will humanoid robot manufacturing become like “playing with building blocks”?

The most amazing thing about the "big factory" is that it fully integrates unit technologies such as intelligence, mechanism, components, control and decision-making to quickly generate various humanoid robot systems, laying the foundation for the self-production, self-construction and self-evolution of humanoid robots.

The "big factory" for designing and assembling humanoid robots is based on the original innovation accumulation of high-precision operations in the "environmental attraction domain" and brain-like intelligent robot theory, and has independently made breakthroughs in core technologies such as high-explosive integrated joints, AI-enabled design, large robot models, and humanoid compliant control.

It is like a large model filled with robot manufacturing knowledge and principles. You only need to tell the large model what kind of humanoid robot you want, and it will generate the best design and manufacturing plan based on your needs and usage scenarios. At the same time, with the support of receiving massive feedback data, the "big factory" will continue to optimize the robot.

The robot system of the "big factory" is integrated with the Zidong Taichu multimodal big model, which was independently developed by the Chinese Academy of Sciences and officially released at the Artificial Intelligence Framework Ecosystem Summit on June 16, 2023. The 2.0 version of the Zidong Taichu multimodal big model adds video, signal, 3D point cloud and other modal data on the basis of voice, image and text, and has full modal understanding, generation and association capabilities.

One of the major difficulties in using humanoid robots is dealing with complex terrain and resisting unknown interference. With AI-enabled "big factories", we can achieve robust control of batch robots based on human musculoskeletal mechanisms, thereby designing highly compliant, high-precision humanoid robots.

The first batch of products "Q Family" were unveiled

On February 24, 2024, the official Weibo account of the Institute of Automation, Chinese Academy of Sciences, released a video, and the Q series of pedigree humanoid robots developed by the humanoid robot research team of the Institute of Automation, Chinese Academy of Sciences, was officially unveiled!

The Q1 bionic high-dynamic robot adopts an ostrich-like low-inertia structure design, which enables accurate tracking and balance control of the robot's entire body posture. In the video, the robot can calmly perform actions that require high precision, such as archery, pouring water, and charging a mobile phone.

The Q2 multi-terrain adaptive robot achieves adaptive and stable motion in various complex terrains indoors and outdoors. The robot adopts the "finite element analysis + topology optimization" design, and achieves high real-time control through observer kernel cutting. The robot can still maintain its original motion state when facing unconventional terrain.

The Q3 high-burst motion robot realizes the robust control of batch robots and the ability to adapt to different environments. The Q3 robot adopts a safety design that decouples the functional structure from the support structure. Through the integration of knowledge engineering and AI high-precision modeling, it is trained and developed for outdoor scene application needs such as farmland operations and field patrols.

The Q4 brain-like intelligent robot is dedicated to revealing the brain, body structure, and operating mechanism of higher animals, and using robots to reproduce the intelligent behavior of higher animals. It is a "popular science robot". From neural circuits to muscle activation, Q4 simulates human mechanisms, structures, and functions, and achieves highly compliant and high-speed movement of redundant and coupled systems.

The Q5 high-concurrency reasoning robot is a first-person perspective robot embodied intelligent body that expands the physical space execution capabilities of Zidong Taichu's multi-modal large model. The robot imitates humans from the principles of the brain, cerebellum, and central nervous system, giving it human intelligence, action, and coordination capabilities, providing important support for smart factory operations and family life services.