Home // ICAS 2023, The Nineteenth International Conference on Autonomic and Autonomous Systems // View article

A Multi-modal AI Approach For AGVs: A Case Study On Warehouse Automated Inventory

Authors:
Ferran Gebellí Guinjoan
Matthias Hutsebaut-Buysse
Gorjan Radevski
Hugo Van Hamme
Erwin Rademakers
Abdellatif Bey Temsamani
Kevin Mets
Tom De Schepper
Steven Latré
Erik Mannens
Tinne Tuytelaars

Keywords: AI based autonomous systems; Automated Guided Vehicles, multimodal AI; navigation; deep learning; neural networks; reinforcement learning; warehouse monitoring & management system; automated logistics & inventory

Abstract:
We present a multi-modal AI approach for Automated Guided Vehicles (AGVs) to perform autonomous warehouse inventory monitoring. A vision module detects and tracks the goods and registers them to the inventory when the confidence score is high. Moreover, we use uncertain detections to direct the AGV to better viewpoints that could lead to new inventory counts. Navigation is done with a Reinforcement Learning (RL) agent trained to perform directed exploration in previously unseen warehouse settings. Because there is not a pre-defined route, we implement a robust way to merge detected items to avoid double counts. We also use speech as an easy way to provide instructions to the AGV.

Pages: 25 to 33

Copyright: Copyright (c) IARIA, 2023

Publication date: March 13, 2023

Published in: conference

ISSN: 2308-3913

ISBN: 978-1-68558-053-7

Location: Barcelona, Spain

Dates: from March 13, 2023 to March 17, 2023