Stony Brook University researchers are teaching AI how to sort your recycling

Researchers at Stony Brook University are at the forefront of an innovative project aimed at enhancing recycling efficiency through artificial intelligence. As anyone who has dealt with a greasy pizza box knows, the recycling process can be fraught with confusion. These items, which may appear to be recyclable, often contain residues that can contaminate entire batches of materials, leading to more waste ending up in landfills. The initiative, which began in January 2025, seeks to leverage AI technology to analyze and categorize municipal solid waste more effectively than traditional methods. With the U.S. being one of the highest per capita waste producers globally, this research could have significant implications for environmental health. The project reflects a growing trend across the nation, where scientists and engineers are increasingly turning to AI to improve recycling systems and waste management. Ruwen Qin, an associate professor and the lead investigator of the project, has actively engaged with local material recovery facilities to understand their challenges. "Collaboration with local facilities is crucial for our research, as their data is essential for developing algorithms that can accurately sort waste," she explained. During site visits, her team utilized low-cost cameras to collect video and audio data, which informed the development of their AI model. This AI system is designed to identify various materials, including paper, plastics, food waste, and textiles, while also estimating their quantities automatically. Supported by the Stony Brook University AI Innovation Seed Grant, the project has enabled Qin to involve graduate students in this critical research. In collaboration with the university's Waste Data and Analysis Center, the team is focused on sampling and characterizing waste streams to gather precise data on material composition. As AI models are trained, they can streamline the process of identifying and separating recyclable materials from non-recyclables. This could significantly reduce the chances of contamination that leads to waste being sent to landfills. While the project is still in its infancy, Qin aims to produce high-quality data that can contribute to the development of more accessible open-source models for widespread use. Looking ahead, Qin is interested in integrating AI with robotics to enhance the sorting process further. The potential for AI in recycling is already being explored by companies like AMP Robotics in Colorado, which has created an AI-driven robotic system for waste processing, and Greyparrot, a London-based startup with a sorting system utilized across multiple countries. Experts like Aurora del Carmen Munguía-López from the University of Buffalo acknowledge the challenges that lie ahead in scaling these AI systems for practical use in larger facilities. However, they also recognize the positive environmental impacts that could arise from improved recycling rates and reduced greenhouse gas emissions. Qin is committed to ensuring that the AI model developed at Stony Brook is a resource that can benefit society at large, emphasizing the importance of making the technology and data publicly available.