Trash is a problem that affects the world and no matter where you go — the arctic, deserted island, or even in the middle of a desert, there’s a real good chance you’ll likely find some trash littered on the ground. The same goes for large urban areas and densely populated cities, only more so and it’s fast becoming a crisis no matter what country you live in. Recycling efforts alone won’t help slow the amount of trash that’s produced daily, but researchers from the Vishwakarma Government Engineering College in India have designed an AI-based system that might help us get there.
The Automatic Garbage Detection and Collection (AGDC) platform uses artificial intelligence algorithms to identify waste in its surroundings and employs a robotic gripper to remove it. It can do so even if the trash is located among objects that are not considered waste. The AGDC was designed using a robotic body, which includes a base, gripper arm, and collection container, while the AI was developed using convolutional neural networks. An integrated camera is used to analyze objects in its vicinity and calculates the position of the waste, grabs it, and stores it in the container.
After the identification algorithm identifies a piece of trash, another estimates its distance from the robot, which is then fed to an onboard microcontroller that then moves the robot to its location. Think of it like a primitive, but capable, WALL-E robot. The researchers have built a prototype that’s capable of collecting 100 to 200g of waste before it needs to empty the container; however they hope to push the amount to 2 or 3-kilograms at some point soon.
Go to Source
Author: Cabe Atwell