Home // International Journal On Advances in Internet Technology, volume 12, numbers 3 and 4, 2019 // View article

Smart Shopping Cart Learning Agents

Authors:
Dilyana Budakova
Lyudmil Dakovski
Veselka Petrova-Dimitrova

Keywords: smart shopping cart learning agent; machine learning; reinforcement learning; Q learning; decision tree; identification tree; ambient intelligence; holographic technology; beacon-based technology; assistive technologies.

Abstract:
The paper describes the design, implementation and user evaluation of utility-based and goal-based intelligent learning agents for smart shopping cart. In keeping user’s shopping habits or user’s shopping list, they guide visitors through the shops and the goods in the shopping center or according to new promotions in the shops, respectively. It is envisaged that concrete implementation of the shopping agents will be running on each shopping cart in the shopping centers or on holographic displays. The k-d decision tree, the best identification tree, and reinforcement-learning algorithm are used for agents learning. The task environment is partially observable, cooperative, deterministic, and a multi - agent environment, with some stochastic and uncertainty elements. It incorporates text-to-speech and speech recognizing technology, Bluetooth low energy technology, holographic technology, picture exchange communication system. Machine learning techniques are used for agents modeling. This kind of intelligent system enables people with different communication capabilities to navigate in large buildings and in particular to shop in the large shopping centers and maximize user comfort. Some initial user opinions of the shopping cart agents are presented. Different embodiments of the shopping agents are discussed like holographic agent embodiment, embodied virtual agent or social robot embodiment. Some approaches of realization of a smart robotic shopping cart that can follow the user are discussed too. The performance of the Q learning algorithm with an introduced environment measures model (a model of the environment criteria) is proposed and explored. Study of the learning parameter is presented. Smart Shopping Cart Learning Agents modeling and development task allows for applying and improving the learning algorithms.

Pages: 109 to 121

Copyright: Copyright (c) to authors, 2019. Used with permission.

Publication date: December 30, 2019

Published in: journal

ISSN: 1942-2652