Skip to content
YUNENTROPY INTELLIGENCE  Product knowledge for global industrial buyersProducts · Suppliers · Sourcing Guides
YunentropyIndustrial Intelligence
Talk to us

Embodied Intelligent Technology Company: Post-90s technology talents take the lead

Tech professionals born in the 1990s are playing a pivotal role at an embodied AI training facility in Guangzhou's Haizhu District. They operate robots to conduct training and drive the advancement of embodied AI technology, making up more than 70 percent of the company's technical workforce.

On the occasion of the May 4th Youth Day, General Secretary Xi Jinping wrote back to representatives of the winners of the Chinese Youth May 4th Medal and New Era Youth Pioneer Award, pointing out that this year is the beginning of the “15th Five-Year Plan” and it is the right time for young people to make contributions.

During the May Day holiday, at the Guangdong Provincial Embodied Intelligent Training Ground in Haizhu District, Guangzhou City, various robots are undergoing immersive training under the operation of young architects and engineers.

Robot training scene

The robotic arm accurately grasps car bolts to complete sorting, the inspection robot autonomously detects hidden dangers along the circuit pipeline, and the humanoid robot completes precise operations… On the front line of scientific and technological innovation, Chinese youth in the new era write a youth without regrets with hard work and responsibility.

“Our task today is to simulate a supermarket scene and let the W1 robot pick up and place common bottled drinks in the supermarket.” Deng Yuici, the basic engine architect born in the 1995s, told Song Lei, the remote control operation engineer of the robot born in the 1990s.

Take a deep breath, the keys vibrate, and your arms move.
Song Lei received the task, put on a VR (virtual reality) headset, and started operating.
As Song Lei moves his arms, the arms of the W1 robot also swing.

“The speed should not be too fast, and the trajectory should be kept smooth.” Deng Yuici warned, and Song Lei controlled the VR handle to make the robot slowly approach a mineral water bottle.
The robot arm pauses briefly when it touches the water bottle. Within one second, the robot senses the state of interaction with the object and records relevant data.

In Song Lei’s eyes, there are many types of robots, but they all have a tacit understanding of each other like good friends. “When operating, you can imagine yourself as a robot, so that you can achieve the state of ‘human and machine integration’. Keep your body muscles relaxed and think about the details of the movements.”

Deng Yuici stared at the sensor and paid close attention to the movement data of the end of the robot arm and the joints.
“The control of speed is very important. When picking up beverage bottles made of different materials, the speed and strength must be slightly different.” Deng Yuici said that the more accurate the movements are and the more precise the cooperation with the robot is, the more accurate and comprehensive the data collection will be.

After the remote control operation is completed, the collected data set is transmitted to the server.
“The accuracy and comprehensiveness of the data need to meet very high standards. For example, in the retail hyper-scenario being tested, robots need to be trained to identify beverage bottles of different materials and various types of shelves, and collect data separately.” Deng Yueci said that the amount of data collected at one time is not enough. New data needs to be generated through simulation amplification. When the data reaches more than 100,000 levels, the next step – model training can be entered.

During automated data collection, Song Lei noticed that the data collected by the robot was in the original format, which was not convenient to read and use. How to optimize it?
Deng Yuici proposed to rewrite a program to convert the original format into a common format to improve the effectiveness of the robot in collecting data in real operating scenarios.

From carefully and repeatedly collecting data to using simulation data to form models, to deploying real machines, the work is hard and interesting.
“It’s a bit like playing a ‘breakthrough’ game. There are so many points to explore.” Deng Yuici said, “The robot has advanced and grown in multiple scenarios, and we are also growing.”

When Song Lei first controlled the robot, his hands were shaking badly, and he was still dizzy after putting on the headset. He often had to get familiar with the field for a long time. “Be careful, do the simplest actions first, such as letting the robot grab a box, and then try new actions.”
Now, whether it is a two-finger clamp wheeled robot in a retail supermarket scene, or a two-finger clamp and dexterous hand robot, Song Lei can operate it skillfully in 10 to 30 minutes.
Even the large-wheeled inspection robots used in complex workshops require a high level of balance on the operator’s body, and Song Lei can operate them with ease.

Despite this, whenever a new robot is delivered to the training ground and unboxed for training, Song Lei still puts himself in the same position as the robot – the “newcomer” at the training ground.
“What kind of work can this robot do? How to make it do a good job?” With thinking, he practiced and summarized his experience day after day to continuously improve his operating capabilities.

Young Engineers and Robots

“The development of embodied intelligence technology requires access to multiple modal data in a large number of scenarios.” Ding Ning, director of the Guangdong Provincial Embodied Intelligent Robot Innovation Center, said that the training ground is equivalent to a laboratory and a training ground. It uses real scene requirements to test operational capabilities and lay the foundation for robots to enter the front line of real production and life in the future.
These young people on the training ground are playing an important role.

“With lofty ideals and determination to work hard, these young people have repeatedly trained and tried in the training ground to transform robots from ‘special tools’ to ‘general intelligence’, and they are also making contributions to the development of future industries.” Gao Fang, Party branch secretary and chairman of Guangdong Embodied Intelligent Technology Co., Ltd., the operator of the training ground, said that among the company’s scientific and technological talents, those born in the 1990s account for 74%.

The majority of young people integrate their personal pursuits into the overall development of the country, constantly create new achievements based on their respective positions, and contribute their youthful strength on the new journey.

Join the discussion

Your email address will not be published. Required fields are marked *

Chat with us Typically replies within a few minutes