Shenzhen Passes China’s 1st AI Regulation; Alibaba Welcomes Former Didi, JD AI Masterminds; Tencent’s Wheel-Legged Robot Develops New Acrobatic Skills
Weekly China AI News from Sep 5 to Sep 11
Dear readers, happy mid-autumn festival! Hope you all enjoyed the full moon with your beloved family and friends. Below is an AI-generated picture of “a bear is eating mooncakes on the moon”. This week, we will discuss Shenzhen’s new special regulation that aims to boost the AI development. Alibaba’s research unit Damo Academy is joined by two top-tier AI scientists who will lead the city brain and XR lab.
News of the Week
Shenzhen Passes China’s 1st Special Regulation on AI Deployment
What’s new: The Shenzhen government last week announced China’s 1st special regulation for AI industries, to take effect on November 1, 2022. Named “Regulations of the Shenzhen Special Economic Zone on the Promotion of Artificial Intelligence Industry 深圳经济特区人工智能产业促进条例,” the Regulations will grant the deployment of innovative AI products and services that haven’t met national or local standard requirements.
Below is a quick summary of the regulation statement:
- The Regulations define AI as “the simulation, extension or expansion of human intelligence by means of perceiving the environment, acquiring knowledge, deducing and deducting and other methods using a computer or its controlled equipment.”
- The “Regulations” require the government to build a public data resource system for the AI industry and encourage and support universities, scientific research institutions, and enterprises to build AI computing centers.
- Shenzhen’s governments and public utilities will take the lead in utilizing AI products and services. The government must also solicit AI solutions for specific application scenarios and attract the world’s leading AI product and service developers.
- As mentioned in the first paragraph, the “Regulations” propose a new system that will widen the public access to innovative products. Based on hierarchical and classified management, the system will approve trial operations of advanced AI products and services whose industries have not yet formulated national or local standards.
Alibaba’s Research Arm Hires Former Didi and JD AI Chiefs
What’s new: DAMO Academy, Alibaba’s research arm, welcomed two new AI scientists last week to remedy its year-long brain drain, Chinese local media Leifeng.com reported exclusively. Jieping Ye is the former VP and chief scientist of KE Holdings (aka Beike), China’s biggest online property agency, and former head of AI Lab at ride-hailing giant Didi Chuxing. Liefeng Bo is the former chief scientist of the Digital AI Lab at China’s largest e-commerce company JD.com. Ye and Bo will serve as the head of the city brain laboratory and XR laboratory of DAMO Academy, respectively. Former leaders of the two labs left this May and August.
Who is Jieping Ye? A top-tier machine learning and data mining expert, Ye is a Professor at the University of Michigan, Ann Arbor. His research interests include big data, machine learning, and data mining with applications in transportation and biomedicine. He has won the best paper awards at ICML and KDD and the NSF CAREER Award in 2010. Ye received a Ph.D. in computer science and engineering from the University of Minnesota in 2005.
At Didi AI Lab, between 2016 and 2019, Ye led the AI development within Didi’s massive transportation platform and used AI to improve order dispatching and fleet management.
Who is Liefeng Bo? Bo is a machine learning and computer vision mastermind who received his Ph.D. from Xidian University in 2007. Before Damo Academy, he was the Distinguished Scientist and the Head of the AI Lab at JD Finance. Before that, he was a Principal Scientist at Amazon, where he led a research team at the cashier-free retail store Amazon Go, which pioneered a new grab-and-go shopping experience using computer vision, deep learning, and sensor fusion technologies. Bo’s research interests include machine learning, deep learning, computer vision, robotics, and natural language processing. He won the National Excellent Doctoral Dissertation Award of China in 2010, and the Best Vision Paper Award in ICRA in 2011.
Tencent’s Wheel-legged Robot Shows New Acrobatic Move
What’s new: Tencent’s Robotics X Lab recently released a new demo of a wheel-legged robot balancing a ball on a flat surface while maneuvering rugged terrains — ladders, cobblestone roads, or ramps. The robot can also walk like a bipedal animal to climb stairs or cross a hurdle.
Named Ollie, the robot made its debut in 2021 and wowed the Internet by showing off with a 360-degree backflip. It can perform multi-tasks from managing stairs to delivering coffee.
Technical details: To grant Ollie a human-like “touch” sense, Tencent researchers added a “skin” to the robot — a tactile sensor mounted on top that can pinpoint the location and weight of any object. The paper Large-Scale Integrated Flexible Tactile Sensor Array for Sensitive Smart Robotic Touch offered more details about the flexible tactile sensor array, which can even detect the pressure of a honeybee.
The robot was also trained to keep balancing in different environments using reinforcement learning and adaptive dynamic programming (ADP), a type of advanced control technology. More technical details can be found in IEEE Robotics and Automation.
“Fengshenbang,” or Investiture of the Gods, is a list to invest the gods of heaven, originated from a Chinese classic myth novel in the 16th century. Last week, Shenzhen-based research institute International Digital Economy Academy (IDEA) introduced an open-source project named Fengshenbang, which encompasses large pre-trained models, user-friendly APIs, benchmarks, datasets, and others. The research team aims to “re-evaluate the open-source community of Chinese pre-trained large-scale models, prompting the development of the entire Chinese large-scale model community.” Read more at Fengshenbang 1.0: Being the Foundation of Chinese Cognitive Intelligence.
Have you ever imagined converting a 2D portrait into a 3D animated character? Researchers from Shanghai Jiao Tong University, Microsoft Asia, and the University of Science and Technology of China proposed a 3D portrait generation network that can create 3D consistent portraits while being controllable according to semantic parameters regarding pose, identity, expression, and lighting. Researchers said their method outperforms prior arts in extensive experiments, producing realistic portraits with vivid expression in natural lighting when viewed from a free viewpoint. Read more at Explicitly Controllable 3D-Aware Portrait Generation.
Tusk Robots (塔斯克), an intelligent pallet robots developer, has raised hundreds of millions of RMB in three financing rounds. Founded in 2020, the Hong Kong-based startup delivers pallet robots widely applied in the fields of medicine, electronics, auto parts, tobacco, the chemical industry, supermarket, shoes and clothing, and eCommerce.
HPY Sorting (好朋友科技), an intelligent ore sorting company, has raised almost RMB100 million ($14.45 million) in its Series B financing round. Founded in 2015, the Ganzhou-based company applies computer vision technology to ore sorting, The company drives progress in the utilization of AI in the mining industry, improving efficiency and making the ore sorting process more environmentally friendly.