LeapMind: Cloud is old! ? : 2020 is the ‘edge’ era
The limitations of cloud computing:
It is also said that data is modern oil.
Data utilization is progressing in various business fields.
Amazon.com: “AWS“,
Microsoft: “Azure“,
Google: “Google Cloud Platform”
The technology for processing large amounts of data in large data centers, called “cloud computing,” has become widespread.
The era of edge computing:
Now, however, the development of “edge computing” has been rapidly expanding while utilizing cloud services.
There is a need for a technology that processes data at familiar terminals (= edge terminals).
Smartphones、
self-driving,
Security cameras,
Processing is now required closer to where the data was obtained.
Now Required Technology: Rapid Processing
Edge computing itself is not a new technology.
Now that AI is in full swing, what is needed is a technology that processes “larger data, faster”.
self-driving:
Instantly analyze “obstacle data detected by car sensors and cameras”
Immediate judgment, such as “stepping on” or “not stepping on” the brakes, is essential.
“Sending video to data center every time for processing” would take too much time.
Drone:
When analyzing the captured video in real time,
Due to limited battery capacity,
Energy saving is essential for AI information processing.
Edge Computing: Market Size
The edge computing market is estimated to grow from $ 170 million in 2017 / 10x in 2025.
Semiconductor chips in the United States and China are leading the world in semiconductor chip development.
Japan Startup: LeapMind
On the other hand, there are also startup companies that are attracting attention in Japan.
It is a startup company LeapMind (based in Shibuya Ward) founded in 2012.
There are only 80 employees.
AI technology: 100 most promising companies
U.S. research firm has announced the 100 most promising artificial intelligence technologies.
One of the “100 Promising Companies Working on Innovative Artificial Intelligence Technology“.
Perform “Processing by deep learning on a palm-sized terminal“.
LeapMind: Main features
“AI algorithm specialized in image processing and semiconductor chip circuit design”
The feature is that image data can be processed with little energy.
LeapMind: Case Study
Efforts for practical use are already underway.
Collaboration with Kawasaki Heavy Industries:
Collaboration with Kawasaki Heavy Industries, which designs train cars.
Analyzing the image taken by the camera installed on the train’s home door with a LeapMind semiconductor chip,
If a foreign object such as an umbrella or clothes is caught in the train door, it is automatically detected and alerted.
It is required to analyze within about 2 seconds from when the door closes until the car departs.
Succeeded in detecting only foreign matter with a probability of 90% or more.
So far
Toyota and
In addition to Mitsui & Co.,
US major IT Intel, etc.
It has received about 5 billion yen from Japan and overseas.
He hopes to expand his business in the United States.
NHK News