Home Artificial Intelligence GIST promotes development of AI semiconductors that mimic insect optic nerves

GIST promotes development of AI semiconductors that mimic insect optic nerves

1
GIST promotes development of AI semiconductors that mimic insect optic nerves

(Photo = shutterstock)

Gwangju Institute of Science and Technology (GIST, Acting President Park Rae-gil) is developing neuromorphic semiconductors using ‘blue technology’ inspired by the principles of nature.

GIST was chosen because the host organization for the topic of the ‘Compound Eye Imitation Neuromorphic Semiconductor for Supervision AI’ promoted by the Ministry of Science and ICT, securing a complete budget of KRW 6.875 billion, including national and native expenses, to develop a synthetic optic nerve imitating insect It was announced on the 18th that it should promote the event of intelligent (AI) semiconductors.

Neuromorphic semiconductors are a essential technology to advance AI-based sensor technologies comparable to lidar, radar, cameras, processors, and memories for future vehicles.

Existing ‘von Neumann structure’ semiconductors have a program-embedded computer structure consisting of three stages: a fundamental memory device, a central processing unit, and an input/output device. This structure executes the listed commands sequentially. On this process, problems as a result of bottlenecks in transmission circuits may occur.

A compound eye imitating neuromorphic semiconductor that mimics the compound eye neural network of insects.  (Photo = Gwangju Institute of Science and Technology)
A compound eye imitating neuromorphic semiconductor that mimics the compound eye neural network of insects. (Photo = Gwangju Institute of Science and Technology)

However, semiconductors that imitate the compound eye neural network of insects are optimized for location and motion detection and might be equipped with low power and high computational power. Because of this, compound eye-mimicking neuromorphic semiconductors might be made right into a sensor-memory-computing integrated system.

This makes it possible to beat the constraints of existing self-driving automotive cameras. It is known as ‘supervision’ because it might probably be developed right into a structure that predicts future movements through the insect’s motion detection ability.

GIST divided its core technology development strategy into three stages. ▲Insect compound eye imitation lens technology development ▲memory-based optic nerve neuromorphic chip development ▲motion detection function implementation and evaluation so as.

To this end, 45 experts in master’s and doctoral level research will likely be assigned. As well as, the Korea Institute of Energy Engineering supports device tests and the Korea Photonics Technology Institute supports optical lenses and device-related test and evaluation equipment.

The aim of this project is that Gwangju’s fundamental industries have modified to ▲mining, clean diesel, home appliances, artificial intelligence, and semiconductors in accordance with the federal government’s stance over the past 20 years, and ▲the ‘crisis level’ of the mining industry is high despite the investment of 1 trillion won within the national budget. ▲Gwangju-Jeonnam’s activities to draw ‘AI semiconductor specialized complex’ are intensifying, and ▲ the necessity to nurture next-generation food is increasing.

Professor Song Young-min said, “There was a case of studying technology that partially copied the visual structure of an animal overseas, however the research we’re conducting was motivated by the concept of ​​digging up and copying the whole insect nervous system.” “Insects have brain cells in comparison with humans. It is far less, nevertheless it is predicted that it should give you the chance to predict the trail of the thing and move it to the specified position with low power while having the advantage of sunshine weight.”

Reporter Hojeong Na hojeong9983@aitimes.com

1 COMMENT

LEAVE A REPLY

Please enter your comment!
Please enter your name here