Introducing an autonomous warehouse with multidimensional movement patterns and collision detection for AGVs

Abstract

The increasing demands for speed, efficiency and accuracy in the trade, production and service sector calls for the need for automation. Automating warehouses and op- timizing last mile delivery can have a plethora of benefits which includes lower labour costs, better space utilisation and quicker material movement. Autonomous Warehouse and Last Mile Delivery (AWARD) as developed by Bright Cape, is a project that uses Artificial Intelligence and Machine Algorithm to offer a logistic platform that can be used for the automation of warehouses. The algorithms developed are used for the coordination of the AGVs (Automated Guided Vehicles) for the movement of goods (pallets) within the warehouse and for the movement and coordination of the drones which can be used for the delivery of the goods to their destinations. The whole system consists of several components such as Warehouse Management System (WMS), Planning and Scheduling module (P&S), AGVs and Drones. The WMS gives information about the identity, position and characteristics of the pallet and also information about when and where the pallets must be moved to. The P&S module takes in this set of orders, assess them and after checking the availability of AGVs, drones and human operators, it develops an optimal algorithm which controls the movement of AGVs within the warehouse. It also plays a role in the navigation of the drones for the last mile delivery. This thesis uses the idea of the AWARD platform and the Light Detection and Ranging (LiDAR) technology used in their AGVs to propose a new system. The work done in this thesis includes proposing an efficient, multidimensional AGVs which can navigate in all directions in the warehouse. The system incorporates machine learning algorithm to detect the presence of obstacles during its movement along the aisle. It can recognize the presence of other AGVs or humans and control its movement so as to avoid collision. It also includes the creation of a database and web application to be used as the warehouse database management system. This work demonstrates the software simulation using Simulink (developed by MathWorks) for presenting the software simulation of the robot and to illustrate its movement along the aisles. Due to the limitations in the hardware implementation, the physical implementation of the system is not proposed in this work.