Applying Artificial Intelligence and Computer Vision to Open Source Distributed Recycling and Additive Manufacturing of Waste Plastic

人们会通过回收来赚钱。在为易拉罐和瓶子提供现金的地方，金属和玻璃的回收已经取得了巨大的成功。可悲的是，回收塑料的现金奖励较少，只有9%的塑料废物被回收利用。其余的则污染了垃圾填埋场或环境。但是现在，一些技术已经成熟，人们可以通过3D打印将废塑料直接回收成有价值的产品，而成本只是正常成本的一小部分。人们用自己的回收塑料制作装饰品和礼物、家居和园艺产品、配件和鞋子、玩具和游戏、体育用品和小玩意，这些都是数百万免费设计的。这种方法被称为分布式回收和增材制造，简称DRAM。DRAM的第一步是用肥皂和水对塑料进行分类和清洗。接下来，塑料需要被磨成颗粒。然后使用回收机器人（废塑料挤出机）将颗粒变成3D打印机的长丝，并用于低成本的3D打印机，或者通过融合颗粒直接3D打印。我们的研究表明，DRAM不仅对环境更有利，而且对于自己生产产品的人来说，它也有很高的利润。使用开源设计来生产您自己的产品比购买它们更具成本效益。如果你使用再生塑料，你可以比商业上的同类产品节省99%以上。这是令人兴奋的，然而DRAM是有限的，目前只有精通技术的第一批采用者使用，因为五大挑战：1)没有简单的方法来确定塑料是否可以用于DRAM， 2)没有低成本的方法来粉碎像PET水瓶一样的薄塑料，3)回收机器人从不均匀的废物原料中制造坏丝，4)低成本的3D打印机在精确的数字复制设计时遭受一系列故障，5)大型3D打印机过于昂贵，并且更容易出现故障。本研究计划的长期目标是克服与将塑料和其他废物升级为开源DRAM系统最终产品相关的五个挑战：1)开发一种基于开源熔融流动指数的低成本快速评估DRAM废物的方法；2)开发一种桌面型节能粉碎机，主要用于3D打印，能够将薄塑料容器分解成薄片；3)集成计算机视觉自动反馈系统，控制回收机器人从废料中挤出3D打印长丝，使其更加均匀；4)集成开源人工智能程序和计算机视觉，实时修复3D打印错误；5)使用回收机器人作为悬挂式3D打印机的打印机。所有这些智能DRAM系统将有助于降低DRAM的成本和复杂性，以便每个人，包括那些生活在孤立和偏远社区的人，土著居民或从事军事/人道主义部署的人，都可以轻松地使用废塑料制造他们需要的产品。

People will recycle for money. In places where cash is offered for cans and bottles, metal and glass recycling has been a great success. Sadly, there are fewer cash incentives for recycling plastic and only 9% of plastic waste is recycled. The rest pollutes landfills or the environment. But now, several technologies have matured that allow people to recycle waste plastic directly by 3D printing it into valuable products, at a fraction of their normal cost. People are using their own recycled plastic to make decorations and gifts, home and garden products, accessories and shoes, toys and games, sporting goods and gadgets from millions of free designs. This approach is called distributed recycling and additive manufacturing, or DRAM for short. The first step in DRAM is to sort and wash the plastic with soap and water. Next, the plastic needs to be ground into particles. Then the particles are either turned into 3D printer filament using recyclebots (waste plastic extruders), and used on low-cost 3D printers or it is directly 3D printed by fusing particles. Our research has shown DRAM is not only better for the environment, but it is also highly profitable for people making their own products. Using open source designs to produce your own products is more cost-effective than purchasing them. If you used recycled plastic you can save over 99% from the commercial equivalent. This is exciting, however DRAM is limited, and is currently only used by technically-savvy first adopters because of five challenges: 1) There is no easy way to identify if plastic can be used for DRAM, 2) there is no low-cost method of shredding thin plastic like PET water bottles at home, 3) recyclebots make bad filament from non-uniform waste feedstocks, 4) low-cost 3D printers suffer from a range of failures for exact digital replication of designs, and 5) large 3D printers are prohibitively expensive and fail more often. The long-term objective of this research program is to overcome these five challenges related to the upcycling plastic and other wastes into final products in open source DRAM systems by: 1) developing a low-cost rapid method to evaluate waste for DRAM based on an open source melt flow index; 2) developing a desktop energy-efficient mostly 3D printable shredder capable of rendering thin plastic containers to flakes; 3) integrating a computer vision automatic feedback system to control recyclebot-extruded 3D printing filament from waste materials so it can be made more uniform; 4) integrating an open source artificial intelligence program and computer vision to fix 3D printing errors in real time; and 5) using recyclebots as printers for hanging cable 3D printers. All together this smart DRAM system will help drive the cost and complexity of DRAM down so that everyone including those living in an isolated and remote communities, indigenous or those doing military/humanitarian deployments - can easily use waste plastic to make products that they need.