Pallet Detection and Docking With Autonomous Forklift;

Abstract

Pallet handling operations in factories and warehouses are vital to ensure continuity and increase productivity. These operations using forklifts face some difficulties in terms of time, cost and efficiency. One way to circumvent these problems is to use mobile autonomous robots. However, autonomous robots may also experience difficulties in such operations. One of these difficulties is autonomous docking. Completion of the docking operation in the desired time, working with good precision and not exceeding the determined cost values are still issues in the field of robotics. The main goal of this study is to provide an easily integrated and low-cost solution for problems frequently encountered in the field of robotics, such as docking and pallet detection in autonomous forklifts. The difference of the image processing mechanism, developed for the pallet detection, is that the sensor used is a low-cost depth camera. The planner has been designed in an easily manipulable way and the requiered precision for picking the pallet have been achieved with a cross-track error of ±3 cm or less, and orientation error of ±4° or less. Performance of the developed system has been tested in a simulation environment on a non-holonomic, tricycle forklift. © 2024 IEEE.