- About Us
- News
- Results Released
- Application Scenarios
- Resource Center
- Meeting
Lab Platform Name
Smart collaborative robot lab platform
Shenyang Institute of Automation of the Chinese Academy of Sciences
Intel Corporation
Using artificial intelligence (AI) to accelerate innovation in industrial robot technology has been incorporated into China's national 2017 government work report. Industrial robots are being promoted on a large scale across China. Traditional robots can carry out only basic movements and are often used as isolated production components, lacking the ability to interact with other operation devices. The industrial field poses increasingly high requirements on robots. As machines take over the work of people, robots are required to execute tasks in the same way as human beings. They are also required to collaborate with other smart components, creating a complete solution. However, industrial robots are still falling short in terms of intelligence, collaboration, and openness. With Industry 4.0 underway, smart manufacturing requests all production components to support device ad hoc networks, requires smart control over production operations, and promotes smart upgrade of traditional robots. Issues regarding control system design and task processing capability are restricting smart manufacturing. Traditional robots feature only motion control and are unable to collaborate with human beings or interact with working environments. In addition, not enough research has been carried out on advanced control of force and autonomy, as well as on complex techniques and special tracks. It has been proven that traditional industrial robots cannot meet the new requirements for intelligence and multi-sense convergence. Therefore, we must build next-generation industrial robots based on Internet technologies, deep learning, and robot operating system platforms.
Using various AI algorithms and smart processing chips as the core, integrating robots and AI is the key to ensuring intelligence, collaboration, and openness of next-generation industrial robots. Achieving complex techniques and collaborative control requires edge computing to make robots more autonomous executing terminals. Therefore, we need to further integrate innovation resources in the industry and promote the establishment of robot innovation centers to bolster industry innovation capabilities. To lay a solid foundation for the industry development, we also need to eradicate bottlenecks in the industry and focus on improving the quality and reliability of key components of robots.
Currently, traditional industrial robots have the following problems:
●Due to the limited R&D and application of industrial robots, most people do not fully understand industrial robots or the opportunities they present.
●There is a lack of unified standards and a general basis platform across the industry.
●Industrial robots have few interfaces, preventing them from effectively communicating with surrounding sensing devices.
●There are a number of different industrial robot controllers, and there is no universal robot control hardware platform.
●With poor intelligence, traditional industrial robots can only be used as operation tools in open spaces and are not capable of independently making decisions.
●Industrial robots have poor network capabilities. As a result, they cannot be controlled in distributed networks.
Because of a lack of collaborative security, industrial robots must operate in isolated security zones.
To resolve the preceding problems, China has formulated strategies for developing industrial robots and AI. This enhances industry technology and expands service fields and international cooperation. We are dedicated to providing lab verification environments for the next-generation industrial robots, which aim to improve the core cognitive ability and flexible operation capabilities of AI. The following lists some of the supported industrial robot experiments:
Convenient human-machine interactions, standard information model verification, cloud algorithm dynamic deployment, group control, in-depth learning, force control operation, visual perception, and independent decision-making.
As Made in China 2025 gets underway, industrial robots are applied to various digital workshops and smart production lines in smart manufacturing. The automobile manufacturing industry has applied industrial robots to a higher extent than any other field, covering the whole process from production, processing, and warehousing of automobiles and their components. The electronics industry requires industrial robots to implement flexible manufacturing and high-speed production. The metal and mechanical processing industries can apply industrial robots to achieve batch and automatic production, reducing the labor costs and improving the production efficiency and management.
In the robot field, the edge computing platform integrates automatic control, information and communications technology (ICT), and AI to form smart application scenarios, including fetching, processing, assembly, and logistics warehousing management.
The smart warehousing management scenario is described as follows:
Figure 4-1 Smart warehousing management
●Scenario elements:
Location: material area and sorting area in the warehouse
Platform: shelves and stands
Target objects: cardboard boxes, bagged materials, and so on
●Task description: Combine scenario elements to form an application.
An example instruction is: "Help me fetch carton A100 from the warehouse."
1、Elementary difficulty: The robots know all the information, including object location and obtaining method.
-Semantic identification: Robots can understand task information and trigger corresponding motion events.
-Visual navigation: Avoid obstacles in the preset route through mode identification.
-Target identification: Identify the target object based on a pre-stored template.
-Robot control: Pick up the object using the visual system according to the demonstrated capture path.
-Robot group control: Perform multi-robot queuing and path planning to ensure the whole queue runs normally.
2、Intermediate difficulty: Robots know some of the information, including target object characteristics.
-Path planning: Plan an execution path based on the known information.
-Target feature extraction: Automatically generate a target template according to the prompts.
-Robot planning: Select a proper capture path according to the actual position of the target object.
-Task management: Allocate tasks to multiple robots according to the number of current tasks to optimize the system power consumption.
3、Advanced difficulty: Robots are in an unknown environment and form a closed-loop system through semantic understanding, network retrieval, and interactions with people.
-Semantic interaction: Robots learn unknown information through human-machine interactions. For example, the robot can learn the position and shape of the target object.
-Unknown environment map composition: Generate a map about the location environment.
-Network retrieval: Learn some information through network retrieval according to the information meaning and representation.
The lab platform combines sensing capabilities such as industrial robot force and vision with AI algorithms to function as the core system in the industrial domain. The group control and cloud access technologies also enable industrial robots to integrate with other production elements and form a unified management platform, instead of functioning as independent operation components. This improves operation efficiency and facilitates dynamic adjustment and tracking of production operation tasks.
The lab platform combines industrial robots with AI technology to generate an industry standard information model and a unified R&D software and hardware framework. This facilitates the implementation of the factory robot operation tool and the next-generation industrial robot, which uses the knowledge diagram to perform task inference and is capable of deep learning. Lab industrial robots produce the following results:
Convenient human-machine interaction: Based on the progress made in voice identification, natural language understanding, facial recognition, and action identification, it is possible to implement convenient human-machine interaction between industrial robots and human beings.
Standard information model: Establish a standard information model for robots, controllers, and demonstrators to standardize communication interfaces between core devices and basic parameters and allow devices to communicate and interoperate with each.
Cloud access: Deploy the high-speed 5G network to connect industrial robots with the cloud, use Big Data to perform operation tasks using smart robots, and enable robots to access the network for subsequent maintenance and tracking.
Group control: Use a distributed control network and a shared robot control platform to implement robot group control. This solves the issue of robots being out-of-sync when they are separately controlled during task collaboration. In addition, the group control technology can be used to rapidly construct and reproduce application scenarios.
Deep learning: By combining robots and AI technology, the high-performance robot controllers can be used to implement deep learning and optimized control when performing special tasks.
Force control operation: Add force sensors to traditional industrial robots so that they can simulate human beings in guide demonstration, assistant control, and force control polishing. Force perception is a core technology in human-machine collaboration.
Visual perception: Add visual sensors to traditional industrial robots to implement smart identification and picking up, fast sorting, and mobile operation. This technology makes traditional industrial robots more intelligent and applicable to fast sorting, visual guidance, and mobile operation.
Autonomy : Regard industrial robots as carriers of edge computing and perform smart parsing on robot control to implement semantic smart control that simulates human capabilities. In addition, the autonomy of robots combines with their deep learning capability to further extend the application fields of robots.
The system lab platform combines sensing capabilities such as industrial robot force and vision with AI algorithms to function as the core device in the industrial domain. The group control and cloud access technologies also enable industrial robots to integrate with other production elements and form a unified management platform, instead of functioning as independent operation components. This improves operation efficiency and facilitates dynamic adjustment and tracking of production operation tasks.
1、The combination of AI and industrial robots complies with the technology trend in the robot industry and China's national strategic deployment for industrial robots.
2、The combination of industrial robots and force control prepares technology reserves so that the assistant robots can be applied to more fields in the future.
3、Improving the visual identification capability of industrial robots can improve their operation mode and working capability.
4、In the future, most digital factories will use industrial robot group management, which reflects the new production mode of smart manufacturing.
5、Deep learning creates more special techniques, solving the problem in which traditional demonstration cannot follow complex tracks.
6、The autonomous capability of robots as executing terminals verifies and highlights the importance of edge computing.
With the advent of Industry 4.0, the rise of labor costs in China, and the transformation of industries, an increasing amount of manual labor is being carried out by robots. China boasts a huge domestic market, a large quantity of talent reserves, and proactive guidance and support from national policies. All of these factors contribute to our advantages in developing industrial robots. Since 2013, China has become the largest industrial robot market in the world, and this market continues to grow rapidly. There is still much room for the growth of the industrial robot market in China, who's industrial robot usage is currently low.
According to the preliminary statistics of the Ministry of Industry and Information Technology of the People's Republic of China (MIIT), China has more than 800 enterprises engaged in robot production, among which more than 200 are robot manufacturers. Most of these enterprises focus on assembly and Original Equipment Manufacturer (OEM), located at the low end of the industry chain. The industry concentration is low and the overall scale is small. According to International Federation of Robotics (IFR), the annual compound growth rate of installed industrial robots was at least 15% from 2016 to 2018. In 2018, the total sales volume of industrial robots will reach about 400,000 worldwide, of which over 150,000 will be sold in the China. The Chinese industrial robot market will continue to be the biggest driving force for the growth of the global industrial robot market.
We can infer from the preceding data that smart collaborative robots have created a market worth billions of US$ in recent years.
The lab platform combines industrial robots with AI technology to generate an industry standard information model and a unified framework for R&D software and hardware. This facilitates the implementation of the factory robot operation tool and the next-generation industrial robot who uses the knowledge diagram to perform task inference and is capable of deep learning. The cloud platform connects systems to build a robot technology sharing ecosystem covering robot design R&D, algorithm verification, performance testing, and system optimization.
Figure 7-1 Robot technology sharing ecosystem
A healthy technology sharing and interactive R&D ecosystem is formed, in which technological advancements of robot R&D are shared across regions.
Currently, robots are mainly applied to industrial production. However, as robot security is improved, their number of functions are increased, and costs are reduced, service-oriented robots will likely become an important part of smart home. To meet requirements for different robot functions, the robot app market will become second only to the mobile app market in terms of the number of users.
The lab platform provides an environment for developing robot applications by building a unified software and hardware framework for robot R&D. The mobile app ecosystem is used as a reference for establishing the robot application sharing mechanism. In this way, users who are writing programs can share their applications to allow others to download or use these apps for free or for a fee. Robot apps implement classified management based on industrial applications and robot models.
Mr. Miao Wei, minister of the MIIT, delivered a speech at the 2017 China Development Forum. According to Mr. Miao Wei, major innovations are being achieved in China's manufacturing industry. We must fully promote cooperation between enterprises, universities, and research institutes to accelerate the construction of innovation centers involving new materials and robots. We must guide and promote the construction of provincial innovation centers, and carry out research on common key technologies and demonstrate industry applications. All these efforts help to solve problems in terms of common technologies and supply shortage across the industry.
With the strong support of policies, China's robot industry is booming. The acceleration of cultivating and developing emerging industries has been recorded in China's 2017 national government work report. AI technology has appeared in the work report for the first time, indicating its significance to national strategies. Because robots are backed up by AI, the upgrade of products in the robot industry is expected to benefit from AI development.