Development of real-time automatic sorting system for color PET recycling process

Pollution from discarded plastic has become a serious environmental problem. The Great Pacific garbage patch consisting of abandoned plastics is killing marine life. In addition, micro-plastics decomposed by solar UV radiation and waves can accumulate in the human body. Recycling plastic is accordingly a critical element of waste management. As part of the solution to this problem, factory automation in recycling plants to handle more waste faster is essential. The amount of reproduced raw plastics is proportional to the inlet speed of the plastics waste stream into a recycling process line. Furthermore, the quality of recycled products with reproduced raw plastics depends on the sorting purity through the line. Thus, an automated system should be capable of real-time classification of the plastics category and rapid manipulation for removing selected plastics. We propose a real-time sorting system for mixed color plastics by applying a machine learning algorithm and a parallel manipulator with a vacuum suction pad. The learning data and picking test samples were collected from a municipal waste disposal site at RM corporation factory. This work shows the feasibility of real-time plastics recycling automation and a future development direction.