Design of Vehicle-mounted Loading and Unloading Equipment and Autonomous Control Method using Deep Learning Object Detection

Abstract

This paper describes a vision system to detect the 3D position of pallets for autonomous forklift vehicles. An accurated image segmentation method based on colour and geometric characteristics of the pallet is proposed. Moreover, the application computes the 3D position and orientation of the pallet and generates the vehicle trajectory to fork it. The system has been tested and experimental results are shown.