Robotic sorting and artificial intelligence in material recovery facilities: a review of published research and expert perspectives.

Material Recovery Facilities (MRFs) are essential to municipal recycling infrastructure, but face difficulties in sorting due to the growing complexity of recyclable material streams and contamination in their inputs. The limitations of traditional sorting technologies and reliance on manual sorters at MRFs have driven interest in integrating robotic systems that combine artificial intelligence (AI) detection with robotic actuation. This paper presents a review of AI and robotic sorting at MRFs, based on insights from recent literature and expert consultations. The literature review uncovered advances in AI algorithms and datasets, application of new sensors and multi-modal sensing systems, and novel robotic gripping and grasping strategies. Expert consultations identified a shift toward deploying AI systems as standalone detection tools, particularly for quality control and facility monitoring, and a movement away from robotic integration due to limitations in speed, reliability, and gripper effectiveness. The analysis reveals three key challenge areas: (1) practical deployment and economic viability of robotic sorting systems, (2) limitations in AI performance and data availability, and (3) robot-specific challenges related to gripping and grasping. The paper outlines future research directions to address these challenges and advance the field.