Wasit Journal of Computer and Mathematics Science, Journal Year: 2024, Volume and Issue: 3(4), P. 40 - 53

Published: Dec. 30, 2024

Context: In the age of rapid e-commerce growth; Robotic Mobile Fulfillment Systems (RMFS) have become major trend in warehouse automation. These systems involve use self- governed mobile chares to collect shelves as well orders for deliveries with regard optimization task allocation and reduced expenses. However, a manner implement such systems, one needs find enhanced algorithms pertaining resource mapping planning movement robots sensitive environments. Problem Statement: Despite RMFS certain challenges especially when it comes distribution tasks overall distances that employees cover. Objective: The main goal this paper is propose new compound model based on RL-ACO optimize RMFS’s assignment navigation. Also, direction study investigate how methods can be applied real-life automation effective large scale. Methodology: This research introduces selection which integrates reinforcement learning Ant Colony Optimization (ACO). Specifically, real gym environment was created perform order training way robotic movement. Reinforcement Learning (RL) models were trained Proximal Policy (PPO) improving dynamic control ACO used computing optimal shelf trajectories. performance also measured by policy gradient loss, travelled distance time taken complete tasks. Results: proposed framework showed potential enhancing efficiency required travel involved. each RL shortest paths identified best route determined total 102.91 units. other values as, value function loss convergence iterations. To build global solution, integration went step forward enabling through combinatorial problems solving. Implications: offers practical, generalizable flexible approach improvement operations thinking

Language: Английский