Toshiba: Logistics support robot “CarriRo”: World’s highest accuracy
-AI that estimates the shape of luggage from visible light camera images from “points”-
November 30, 2020
Toshiba:
From images taken with a general visible light camera
Even in individual areas of irregularly stacked objects
AI (artificial intelligence) technology that estimates with high accuracy
Announced that it was developed.
Picking robot experiment:
We have released a demonstration experiment using a dataset for picking robots.
Measurement error in estimating the object area,
Beyond “combination of existing AI technology and 3D sensors”
It is said that it has achieved “the world’s top performance” (the company).
Developed AI technology:
Use a visible light camera that requires very little adjustment.
“The shape and recognition accuracy of the luggage is equal to or better than that of the 3D sensor”.
“Point candidate-based object area extraction technology”:
Estimate the most probable candidate point from the pixel,
The “method of estimating the area of the object including the candidate point” was adopted.
Therefore, the name is also “Point candidate-based object area extraction technology”.
Launching unloading robots:
In addition to the unloading robots for distribution warehouses scheduled to be launched in 2021,
Factory marshalling and boxing,
Unmanned cash registers at retail stores
It is also expected to be applied to inventory.
-MONOist
https://monoist.atmarkit.co.jp/mn/articles/2011/30/news040.html
[Toshiba] Developed AI that estimates the area of individual luggage with the world’s highest accuracy from images from ordinary cameras –CNET Japan