This self-driving bike shows progress in China’s chip design capabilities

Chinese researchers have developed a new hybrid chip architecture that can take the world one step closer to achieving General Wisdom Artificial (AGI or artificial general intelligence) and a future full of decks Thinking machine like human. And it also shows progress in China’s advanced chip design capabilities.

To demonstrate new advances in their chips, researchers created a self-driving bicycle using this new hybrid chip. Combined with other parts such as cameras, gyroscopes, speed gauges, operating motors and steering wheel controls on a new bike or chip – called Tianjic – helps the car not only self Drive but also can walk around obstacles, even respond to simple voice commands.

Tianjic is a result of research by Professor Shi Luping and colleagues at Tsinghua University, China. According to the researchers, Tianjic is designed by integrating computer science-based approaches and neuroscience-oriented approaches to developing AGI.

This is a special feature of Tianjic’s design architecture, when it is a combination of Von Neumann’s design for conventional chips and artificial neural network design for designed processors. for machine learning.

To do this, Tianjic has a multi-core architecture, with multiple upper blocks that can be reconfigured and a seamless data path with hybrid coding schemes. Therefore, Tianjic is not only compatible with machine learning algorithms in computer science, but it is also easy to implement human-inspired algorithms, and many other coding schemes.

Thanks to the combination of these two architectures, Tianjic can run conventional software, as well as operate artificial neural networks so that the bike can perform tasks such as detecting obstacles. objects, steering control and self-balancing, and voice recognition, via vehicle-mounted sensors.

In an interview last Thursday, Mr. Shi, a professor at Tsinghua University’s Brain Research Center, said: ” The chip for this General Intelligence is potential. It can be applied in all different industrial sectors, and self-driving cars, robots and automation will be the areas where this chip can make a difference . ”

Shi said that the chip clearly demonstrates “its ability to support many different code schemes and be adaptable in complex environments .”

Self-driving bicycle equipped with Tianjic and other sensors.

According to Shi, efforts are now aimed at cultivating startups focused on developing Tianjic chip-based applications of this study. Their goal is to launch a version of the chip capable of mass-producing early next year, but without giving details of the potential trading partner, as well as the cost of such a plan. .

The chip also shows China’s progress in developing chip self-design capabilities. So far, China is still struggling to build its own chip industry, and the weakness of this technology capacity has been exploited by the United States in the ongoing trade war.

Although the world’s most advanced self-produced computer chips are still out of reach, Chinese researchers are showing they can make dedicated AI chips of the same quality. any country

Refer to Technology Review