In recent years, the development of embodied intelligence has been paid more and more attention in the field of artificial intelligence.
Embodied intelligence emphasizes the interaction between the agent and the physical environment, and realizes more natural and efficient interaction by simulating the perception and action ability of the organism.
With the progress of sensor technology and the improvement of computing power, the application prospect of embodied intelligence is broad, from service robot to virtual reality, and then to self-driving car, its potential is gradually released.
On December 18, 2024, the second embodied Intelligent Industry Development Forum in 2024, hosted by Galaxy Automobile, came to a successful conclusion.
The forum focused on humanoid robot shape design, embodied intelligence, motion control, human-computer interaction, cost-effectiveness, software algorithms / artificial intelligence, key technologies and trends, application scenarios and other hot topics.
Many industry technology celebrities and authoritative experts share topics around the topic, which provides innovative ideas for promoting the practice and development of the embodied intelligence industry.
This conference thanks Gates Electronics, Zhenmaojia Semiconductor, Arctic University, Zhiyuan Electronics, same Star Intelligence, Sihui Fushi, Zhisheng Software, Zhinan Technology and Junpu Intelligence for their strong support! The organizers welcome the speech, and Zhou Xiaoying, the CEO of Galaxy, delivered a welcome speech as the organizer.
As we know, “embodied intelligence” has risen rapidly in recent years and has become the focus of attention in the field of science and technology.
From academic research to industrial practice, people from all walks of life are actively exploring the broad application prospects and great potential of embodied intelligence.
As the world’s leading information service platform for the automobile industry, Galaxy always follows the development trend of the industry, devotes itself to promoting the healthy development of the industrial chain, and provides professional information services for the automobile industry.
In today’s forum, we have invited leaders from academia and the business community to discuss the application of embodied intelligence in the automotive industry, technological breakthroughs and future development opportunities.
We look forward to more cooperation and innovative thinking through in-depth discussions and exchanges between guests and colleagues.
Zhou Xiaoying | CEO of Galaxy Automotive, embodied the data bottleneck in intelligence.
Gao Yang, assistant professor of Cross Information Research Institute of Tsinghua University and co-founder of thousand fathom Intelligence, shared the main topic of “data bottleneck in embodied intelligence”.
At present, the rapid development of AI technology has reached an amazing ability and innovation, which is supported by massive data, that is, the big language model.
For robots, there is still a lot of room for development in the current data support.
We believe that strong enough data support is the key point for embodied intelligence to achieve high-level development.
Therefore, we focus on the research of data acquisition equipment to solve the problem of data planning.
Through the interaction between the robot and the environment, strengthen the learning process, obtain information and feedback to itself, so as to constantly optimize the learning process, and then form a data flywheel.
The learning ability of robot is a necessary condition for the formation of data flywheel, which is particularly important for the development of embodied intelligence.
It is expected that ten years later, with the advent of the aging society, humanoid robots can become important assistants in people’s daily life.
Gao Yang, Assistant Professor of Cross Information Research Institute of Tsinghua University, co-founder of thousand fathom Intelligence, in-depth analysis and prospect of humanoid robot industry chain.
Xue Sheng, an analyst at Galaxy Automotive Research Institute, mainly shared the in-depth analysis and prospect of humanoid robot industry chain.
Humanoid robot accelerated commercial landing from 1973 to 2021, and the technology gradually matured.
Starting from 2022, Tesla and Xiaomi will enter the bureau to promote market competition, and AI and cloud technology will promote intelligent development in 2024.
It is expected that from 2024, the humanoid robot industry will enter a period of rapid development, which is mainly driven by the continuous progress of technology and the growth of market demand.
Intelligent industry policies are favorable, and robot industry policies are issued all over the country, focusing on technological breakthroughs, industrial chain improvement and application promotion, and strive to form large-scale industrial clusters and cultivate core enterprises by 2025-2027.
The support of external policies will effectively promote the development and application of robot innovation.
In the future, with the iteration of technology, the application scene of humanoid robot is divided into three stages, the first stage reaches the application standard of “replacing repetitive labor”, and the second stage begins to move towards more scenarios of all kinds of manufacturing factories and business services.
the third stage of embodied intelligence has become a reality.
At present, the difficulty of development in the industry lies in algorithm, specification and cost.
Its successful application requires powerful universal algorithms and industry standards to solve privacy and security problems.
at the same time, cost reduction is the key to machine replacement.
Xue Sheng | analyst of Geshi Automotive Research Institute, possessing intelligence, opened the way to AGI.
Huang Dongyan, corporate mentor of Shenzhen International Graduate School of Tsinghua University and visiting researcher of Integration Institute of Shenzhen Advanced Technology Research Institute of Chinese Academy of Sciences, gave a speech with the theme of “embodied Intelligence, opening the way to AGI”.
Embodied intelligence is a key path to general artificial intelligence (AGI).
Embodied intelligence not only plays an important role in the control and perception of robots, but also similar to the cerebellum of human brain, which improves the level of intelligence through interaction with the environment.
The cooperation between cerebellum and cerebellum is the core issue of the development of machine intelligence.
The progress of embodied intelligence will accelerate the realization of this cooperation mode and lay the foundation for the future development of AGI.
, Huang Dongyan | Enterprise mentor of Shenzhen International Graduate School of Tsinghua University, visiting researcher of Integration Institute of Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences, round table forum stage, Zhou Xiaoying, CEO of Galaxy, as host, enterprise mentor of Shenzhen International Graduate School of Tsinghua University Huang Dongyan, visiting researcher of Integration Institute of Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences.
Shen Tianyao, CTO of Junpu Institute of artificial Intelligence and humanoid Robotics.
Yu Lei, co-founder of Xinghaitu.
Shi Qingfan, founder and chief executive officer of Shanghai Youth Tour.
researcher and doctoral supervisor of National key Laboratory of Multimodal artificial Intelligence system, Institute of Automation, Chinese Academy of Sciences Wang Peng, founder of Sinoke Silicon era, and Chen Ming, vice president of the second Department of Equity financing of Huaan Securities Co., Ltd., as discussion guests, conducted in-depth discussions on such topics as “embodied intelligent situation development, landing speed, development direction, specific commercial landing practice and development bottleneck”.
Shen Tianyao, CTO of Jupu Institute of artificial Intelligence and humanoid Robotics, mainly discusses the engineering landing of embodied intelligent robot.
The engineering landing challenge of embodied intelligent robot, including cost, large-scale product inventory, software algorithm and so on.
Breakthroughs are still needed in type A training, data acquisition and processing.
At present, the product form of humanoid robot is not mature and needs to be optimized according to specific needs.
Clear demand is the key to promote product development.
In the field of embodied intelligence development, the lack of clear demand orientation in product development is due to the lack of clear scenarios.
Jupu Intelligence is committed to helping technology companies clearly define the characteristics of robot products, from the perspective of mechanical design, to provide a large number of multiple scenarios for enterprise products to do landing verification, and to investigate various unstable factors.
promote a series of practices of product promotion in the future and achieve large-scale applications.
Shen Tianyao | CTO of Jun Pu Institute of artificial Intelligence and humanoid Robotics, research and exploration of swarm intelligence technology for industrial field.
Liu Shengxiang, director of Lailai Automotive prospective Manufacturing Engineering Department, introduced the research and exploration of swarm intelligence technology of Lailai Automobile for industrial field.
The manufacturing industry has experienced the evolution from scale manufacturing, lean production, flexible manufacturing to intelligent manufacturing, and the technological upgrading and iteration of each stage promote the development of the industry.
At present, the shortening of the product launch cycle and the fluctuation of the batch market put forward higher requirements for the flexibility and toughness of the manufacturing system.
Although digital and industrial Internet of things have made progress, there is still a need for a breakthrough in the deep integration of intelligence and manufacturing systems.
The decline of hardware cost has promoted the further development of discrete manufacturing industry, but the phenomenon of data isolated island still needs to be solved.
Lailai Automobile has carried out a full stack technical layout in terms of intelligent attributes, including intelligent cockpit, panoramic interconnection, vehicle global operating system, etc.
, and attaches importance to the basic role of digital operation and artificial intelligence in intelligent manufacturing.
The value of humanoid robot in the field of automotive industry manufacturing will mainly focus on physical trial and error, repeated automation scenarios, etc.
, to solve tasks that require intelligent attributes and multimodal interaction.
In the future, the combination of humanoid robot and industry should be based on the value of practical application scene, solve specific problems and form product generalization, so as to promote the manufacturing industry to a higher level of intelligent transformation.
Liu Shengxiang | Director of the prospective Manufacturing Engineering Department of Lailai Automobile, humanoid dexterous operation embodied intelligent robot, researcher and doctoral supervisor of the National key Laboratory of Multimodal artificial Intelligence system, Institute of Automation, Chinese Academy of Sciences.
Wang Peng, founder of SinoScience, shared the report on the theme of “humanoid dexterous Operation of embodied Intelligent Robot”.
The key elements of robot operation ability lie in both hardware and software, including the design of dexterous hand and the integration of multimodal perception system.
In the aspect of software, planning and decision-making need to be realized through learning algorithms to deal with complex environments and tasks.
At present, robot operation is widely used in the field of industrial services, but the penetration rate is low in home and special application scenarios.
The main challenges it has to deal with include the diversity of the environment and the complexity of tasks, especially in the family environment, objects and tasks within the same family are quite different, corresponding to the need for a more complex programming system.
In order to improve the adaptability of the robot to the scene, it is necessary to combine the software and hardware technology to realize the generalization of diversified objects and scenes.
Man-machine fusion is an important way to realize the general multi-task capability of robot, and complex tasks can be better accomplished through man-machine cooperation.
Data collection and multi-task general algorithms enable the robot to adapt and perform tasks in a changeable environment.
Therefore, data collection is the key to improve the ability of the robot.
The collection methods mainly include generative data and actual scene data, and the future trend will combine the two to improve efficiency and practicality.
Wang Peng, researcher and doctoral supervisor of National key Laboratory of Multimodal artificial Intelligence system, Institute of Automation, Chinese Academy of Sciences, founder of Silicon Age of Chinese Science, key technology of general humanoid robot: data distribution service.
Shi Qingfan, founder and CEO of Shanghai Youth Tour, mainly shared the key technology of general humanoid robot, namely data distribution service.
Juvenile Travel Technology Co., Ltd. is a mainframe factory focusing on the independent research and development of full-size universal humanoid robots, using AI+ robots to create a world-class intelligent “chain master” enterprise with ecological dominance and global competitiveness.
The company’s development is based on the vision of “creating an intelligent helper to serve the better life of mankind”.
DDS (data Distribution Service), as a middleware standard for data communication in distributed real-time systems, aims to provide efficient, reliable and real-time data transmission capability to meet the application scenarios with strict requirements for data transmission delay, throughput, reliability and so on.
Its working principle is based on the publish-subscribe model, in which the publisher publishes the data to a specific topic (Topic), while the subscriber expresses interest in the specific topic and receives the data from the publisher.
This mode can decouple the producers and consumers of data, and improve the expansibility and flexibility of the system.
It can be widely used in intelligent robots, intelligent transportation, intelligent industrial production and other fields.
It can quickly distribute all kinds of data of intelligent robots and self-driving vehicles and feedback the status of vehicles.
In addition, it can also achieve real-time transmission of production equipment data and issue industrial production control instructions, in order to improve production safety, improve production efficiency and quality, reduce costs, and promote the development of intelligence in various fields.
Shi Qingfan, founder and CEO of Shanghai Youth Games, embodied the intelligent end-to-end control base.
Fang Tingting, founder of Guoxunxin Micro (Suzhou) Technology Co., Ltd., shared the theme report of “embodied intelligent end-to-end control base” to the guests.
Guoxun Core Micro is a high-tech enterprise that focuses on the technology research and development, product development, production and sales of industrial real-time operating system and industrial soft PLC, and is committed to autonomous controllable core technology products for intelligent manufacturing.
The company develops independent NOM series products based on ARM architecture chip platform, and combines independently developed industrial real-time operating system NECRO and NECRO SUPER STUDIO automation software to form a complete set of control system platform products, which can provide open automation control systems, automation development tools and complete automation and motion control solutions for various industrial control fields.
At present, the company’s cutting-edge solutions support humanoid robots, special robots, unmanned vehicles, drones and other terminals, which are widely used in manufacturing, logistics, agriculture and other fields.
In the embodiment manufacturing scene, the cooperation of multi-sensors and actuators is realized, and the level of production automation is improved.
At the same time, flexible manufacturing is realized on the maglev high-speed line, which shows the high flexibility and adaptability of the system.
The futureWe will continue to deepen cooperation with ecological partners, promote the application of intelligent control systems in more fields, and jointly explore all-round intelligent solutions of land, sea and air.
, Fang Tingting, founder of Guoxun Core Micro (Suzhou) Technology Co., Ltd., multimodal large model makes embodied intelligence possible.
CEO, chief scientist of iFLYTEK Robot and Anhui General Robot Technology Co., Ltd., and Ji Chao, assistant dean of humanoid Robotics Research Institute of University of Science and Technology of China, shared it with the theme of “multimodal large model makes embodied intelligence possible”.
The competition in the field of AI is becoming increasingly fierce, and the promotion of high-end intelligence is inseparable from the breakthrough of cognitive intelligence model.
The multimodal large model includes a variety of sensory information such as language processing, image, video and so on, which breaks through the limitations of the single modal model, makes the robot have stronger perception and understanding ability, and improves the comprehensive intelligence level of the robot.
The emergence of cognitive large model sets off a new trend of general-purpose robot and promotes the application of robot in various scenes.
Its applications in the robot field include spatial perception, logical relationship understanding and so on, which enable the robot to better adapt to the complex environment and task requirements.
The large model shows great potential in commercial applications, especially in industrial, medical, automotive and other fields.
The future man-machine cooperation mode will be an important direction for the application of large models.
Robots can help human beings to complete the front-end complex tasks, but the final decision-making power is still left to human beings.
Such man-machine cooperative decision-making will be the main direction of the development and application of humanoid robots.
IFLYTEK has made a number of innovations in the field of large models and robots, including the development of multimodal capabilities, the application of reinforcement learning and so on.
Committed to creating end-to-end solutions, through the cloud-edge integrated design to achieve efficient and safe robot applications.
In the future, robots will land in more fields, but they need to deal with data, generalization, ethics and other challenges.
, Ji Chao | Chief scientist of iFLYTEK Robot, CEO of Anhui Leadong General Robotics Technology Co., Ltd., and Assistant Dean of humanoid Robotics Research Institute of University of Science and Technology of China.
The commercial status and possible landing scenarios of humanoid robots are discussed again.
Chen Xiao, founder of robotics and CEO, re-discussed the commercial status and possible landing scenarios of humanoid robots.
2023 is the first year of humanoid robot development.
at present, there is obvious financing polarization in humanoid robot racetrack.
Head enterprises get large amount of financing, but small start-up teams face financing difficulties.
The phenomenon of “unable to buy, no one to buy, unable to sell, dare not sell” in the sales market shows that there is a serious mismatch between market demand and technological maturity.
The future development not only needs to solve the core scene problems, but also needs to promote the breakthrough and application of technology, especially in the field of subdivision.
By solving the problems in specific scenarios, enterprises can accumulate technology and gradually promote commercialization.
Appeal for financing support for small and medium-sized enterprises, which is very important to promote the healthy development of the whole industry.
Chen Xiao | Origin robot founder, CEO, humanoid learning: making robots more like human from the inside out, Cui Yuxiang, a navigation technology expert embodied in Zhejiang humanoid Robot Innovation Center, delivered a keynote speech on “humanoid Learning: making robots more like people from the inside out”.
Zhejiang humanoid Robot Innovation Center has made many intentional achievements in the research and development of humanoid robot technology, especially in the fields of motion control, arm operation, mobile planning and human-computer interaction.
At present, humanoid robots still face many technical challenges, especially in terms of flexibility and autonomous learning.
Through the data-driven approach, Zhejiang humanoid Robot Innovation Center is committed to improving the “humanism” of the robot system.
Pilot 2 NAVIAI effectively improves the humanoid movement, operation and mobility of the robot through technical routes such as mechanism fusion learning, rapid model deployment and continuous learning.
In the future, embodied intelligence technology will greatly promote man-machine integration and bring innovative applications to many fields such as family, education, service and so on.
Cui Yuxiang | Zhejiang humanoid Robot Innovation Center embodied navigation technology expert, dexterous hand-assisted embodied intelligence.
Wang Zhenkun, CMO of Shanghai Aoyi Information Technology Co., Ltd., introduced the company’s cutting-edge innovative products such as dexterous hands.
Proudly started from the electromechanical bracelet project, and gradually transformed into the field of disabled assistance and robot applications.
Since 2015, the company has initially focused on the research and development of electromechanical bracelets for use with VR glasses.
With the technological progress and changes in market demand, the company began to explore new application scenarios, and finally applied technology to the field of disabled assistance and robotics, successfully achieving product diversification and market expansion.
As the company’s flagship product, the application experience of bionic dexterous hand has made remarkable achievements.
After 7 years of market testing, at present, the arrogant bionic dexterous hand has been maturely used in many industries, can achieve accurate and flexible operation in practical applications, and the physical interaction ability of the robot has been significantly improved.
, Wang Zhenkun | CMO of Shanghai Aoyi Information Technology Co., Ltd., humanoid robot matrix product scheme: accelerate the development of embodied intelligence industry.
Chen Wei, product director of Titanium Tiger, introduced the theme of “humanoid robot matrix product scheme: accelerating the development of embodied intelligence industry”.
Titanium Tiger Robot focuses on joint product applications and humanoid ODM business.
The core product is humanoid joint, which aims to achieve high integration and high performance to meet the needs of complex sports.
Products are widely used in industrial, special operations and other fields, such as power grid maintenance, glass manufacturing and automobile steering system.
The company focuses on technological innovation, addresses customer-specific needs through customized solutions, and explores new technologies such as exoskeletons to enhance the user experience.
Under the current industry background, data problems have become increasingly prominent, including the diversity of data collection methods, access and costs, which play a key role in the strategic market.
But at present, the ultimate goal of data is not clear, including data closed loop, management and distribution to the transaction of the whole chain need to be further explored and optimized, in addition, enterprises need to clarify their own development direction, whether hardware or software, should focus on providing core solutions, such as power solutions and industry-specific technical frameworks.
In the field of emerging technologies, enterprises should build a solid foundation and enhance their solution capabilities in specific industries in order to promote the development of new productivity.
Chen Wei | head of Titanium Tiger Robot products, exploring machine vision requirements through human engineering, Fu Bing, general manager of Shanghai Kam Vision Technology Co., Ltd.Kay shared the theme of “Exploring machine vision needs through human factors engineering”.
, Keim is a technology research and development company integrating optical design, automotive camera module assembly, production and sales.
Using Sham’s Law of Austrian Army Captain Sidor Shamvrog a century ago, the core indicators of glass mirrors and the restoration of magnification and magnification aspect ratio parameters are perfectly realized.
The optical design parameters of the CMS/BSD lens are calculated by calculating the relationship between the vehicle three-dimensional data and the human eye observation angle combined with the field of view observation area of “GB15984-2022” and “GB/T39265-2020” regulations.
Through big data model modeling, the angle between the lens focal plane and the CMOS sensor is accurately calculated and the shell is processed using CNC technology.
Structural parts are used to solve the problem of image quality taken by non-forward mounted cameras, which reduces the cost caused by high-code operation of image software.
and power consumption.
At the same time, it is also committed to improving the installation reusability of AI algorithm cameras on different vehicle models, and solving the problem of perspective distortion caused by external parameters changing due to non-forward installation, including magnification and its aspect ratio distortion, MTF (Modulation Transfer Function) Reduction and other core problems that affect algorithm reuse.
The decompression and exhaust/lens cleaning device ensures the safety, reliability and long life of the product throughout its life cycle.
, Kaim has completely independent intellectual property rights and has currently obtained 8 patent authorizations.
Another 2 PCT international patents have been submitted to landing countries, and 4 utility models and 3 invention patents are under review.
The product is used in the AI algorithm image recognition industry that has high requirements on magnification, magnification aspect ratio, and MTF parameters.
, Fu Bingkai| General Manager of Shanghai Kemm Vision Technology Co., Ltd., so far, the 2nd Shaped Intelligent Industry Development Forum has come to a successful conclusion.
The guests on site have contributed many cases and shared cutting-edge views to the development of the Shaped Intelligent Industry, covering everything from technological development to industrial trends.
The multi-faceted perspective provides this forum with rich knowledge content and forward-looking industry insights.
The successful convening of this forum provided a communication platform for technological innovation in the industry, promoted the sharing of the latest scientific research results and application cases, and helped enterprises and scientific research institutions understand cutting-edge trends in the industry.
At the same time, it promotes interdisciplinary and cross-domain cooperation, enhances the synergy effect of technology research and development, and will play a key role in promoting the future development of specific intelligent technologies.
Return to First Electric Network Home>.
