Robot recycling 垃圾分揀機器人

本集內容

學習要點

有關 “technology 科技” 的詞匯

邊看邊答

What percentage of stationary items could the robot identify?

文字稿

This robot can automatically sort recyclable rubbish. The RoCycle system by M.I.T. has a soft, puncture-resistant hand. Pressure sensors on its fingertips detect an object’s size and material. It then autonomously places the item in the appropriate recycle bin.

這個機器人能自動分類可回收的垃圾。麻省理工學院的 “RoCycle” 系統用一隻柔軟的、防穿刺的手來分揀回收垃圾。指尖上的壓力傳感器可以探測到物體的大小和材質。然後它自動將物品放入相應的回收垃圾箱。

Professor Daniela Rus, Director, MIT CSAIL
With computer vision alone, the systems are not able to separate paper from plastic. Many paper and plastic cups look the same, but by introducing the ability to squeeze the objects and to know whether it’s flexible or not – we are able to go one step beyond what today’s methods can do.

丹妮拉·魯斯教授麻省理工學院計算機科學與人工智能實驗室
“單憑計算機視覺，這些系統無法將紙張與塑料分開。許多紙杯和塑料杯看起來是一樣的，但是通過引入擠壓物體的能力，並辨別它是否有彈性、可彎曲，我們則能夠超越現有的方法。”

The goal of the system is to reduce the back-end cost of recycling. It currently has 85 percent accuracy in identifying stationary items, but only 63 percent accuracy on a simulated conveyer belt.

該系統的目標是降低後端回收成本。目前，它識別靜止物體的準確率為 85%，但在模擬輸送帶上的準確率僅為 63%。

A common error was identifying paper-covered tins as paper. But how are researchers looking to improve the system?

一個常見的錯誤是把外包裝由紙覆蓋的金屬罐頭盒識別成紙。那麼研究人員在如何改進這個系統？

Professor Daniela Rus, Director, MIT CSAIL
We plan to create a much more detailed sensorised skin. We plan to develop the hand at different sizes and we plan to improve our algorithms for recognition. We’re very excited to see the use of robot automation in solving a problem that matters globally.

丹妮拉·魯斯教授麻省理工學院計算機科學與人工智能實驗室
“我們計劃創造一個更精細的傳感皮膚。我們還計劃研制不同大小的手，並改進我們的識別算法。我們很高興看到機器人自動化在解決一個全球性問題上的應用。”