Robotic Grasping for Automated Sorting of Complex, Highly Contaminated Industrial Food Waste: A Benchmark Study

Abstract

Food waste management plays a vital role in maintaining a sustainable ecosystem, however, the presence of inorganic contaminants within food waste significantly hinders this potential. Robotic automation offers a promising solution to accelerate waste sorting, yet the diverse and unpredictable nature of contaminants poses major challenges to robotic perception and grasping. This benchmark study explores the feasibility and limitations of conventional robotic grasping systems, replicating real-world industrial conditions to highlight the complexities of food waste sorting. A comprehensive automated robotic grasping pipeline is introduced, integrating advanced 6D grasping pose detection, collision-free robotic arm motion planning, and effective grasping with three top-performing robotic end-effectors. Extensive experimental evaluations (up to 1500 robotic grasps) compare the performance of different gripper designs and the corresponding grasping strategies under three high-fidelity environmental scenes, providing valuable insights into the limitations of the current robotic system. Experiment results demonstrate the significant strengths of each gripper when dealing with objects of varying types or in different environments. This is critical for enhancing robotic sorting capabilities, particularly in advancing multimodal gripper technology.