Decision-Making for Logistics Processes of the Robotic Warehouse Based on Markov Networks

Abstract

Modern warehouse logistics has seen tremendous development and, at the same time, faces many challenges due to the constant volatility of the market and unpredictable demand. At the same time, it is not always easy to coordinate various processes, taking place in a warehouse. In most cases, standard management methods do not yield a positive result, which leads to irrational use of resources and increased costs. This article proposes a way to solve the problem of warehouse robot routing based on a combination of expert systems and Markov logic networks. Markov networks, as a mathematical tool for modeling random processes, help to understand how different states of a system are interconnected. In warehouse logistics, this means that it is possible to predict what state the system will be in in future based on current data and what has happened before. This is very important when you need to make serious decisions at the management level. Another important point is expert systems that combine the knowledge of experienced professionals and Markov networks. This makes it possible to formalize the knowledge of people with extensive practical experience and use it in an automated manner. This combination makes it possible to combine practical observations with mathematical analysis and ensure adaptability. With the help of a combination of expert knowledge and models, it is possible to take a more rational approach to working with warehouse processes. The proposed article discusses the method of route optimization in the face of uncertainties arising in the dynamic environment of a logistics warehouse. Method has been developed that combines explicit expert rules with probabilistic models of logic networks. A conceptual model for the combined system is also proposed, how to convert data from the expert system into formulas for logic networks is explained, and computational aspects are considered. Possible directions for future research, including automatic learning methods and adaptive systems, are discussed. The results of the article will create a basis for intelligent control systems that can solve routing problems taking into account many factors that affect the movement of robots.