Methods of Evaluating 3D Perception Systems for Unstructured Autonomous Logistics

This study introduces methods of evaluating 3D perception systems, such as Time of Flight (ToF) systems, for automated logistics applications in unstructured environments. Here perception is defined as a system’s understanding of its environment and the Objects Of Interest (OOI) within that environment, through hardware consisting of cameras or depth sensors. Current computer guided machinery that rely on perception systems, such as an Autonomous Guided Vehicle (AGV), require structured environments that are specifically designed for such a machine. Unstructured environments include warehouses or manufacturing facilities that have not been tailor designed or structured specifically for the purpose of using a computer guided machine. In this study, two methods are proposed to assess 3D systems proposed for autonomous logistics in unstructured environments. The results of this study indicate that the methods presented here are suitable for future and comparative 3D perception and evaluation in this space.