An algorithmic method for warehouse order picking with congestion and one-way aisle constraints

Abstract

Order picking is a fundamental management process in warehouse management systems and it has a great impact on the cost of operations, efficiency and quality of the services. Although the use of traditional routing and shortest path algorithms is a frequent solution to the problem of order-picking, it frequently ignores real-world constraints like the congestion and one-way aisle patterns, which can significantly impact the viability and effectiveness of routes in the real-world warehouse setting. To overcome these drawbacks, this thesis is concerned with creating and analyzing congestion-aware routing plans in a realistic warehouse environment with order-picking. A simulation based benchmarking model is suggested to compare systematically classical, heuristic, hybrid, and metaheuristic routing algorithms based on unified warehouse layouts with one-way aisle constraints and a travel cost that is a congestion cost. The structure makes it possible to model route execution, traffic congestion, and collision impacts that allow fair and consistent comparison between algorithm methods. Performance variables are calculated on standardized measures, including quality of solutions, planning efficiency, collision impact, resource use and scalability to problems of different sizes. The comparison analysis shows that even though exact routing algorithms only work well in small scale problems, heuristic routing methods are quicker but more vulnerable to congestion impacts. Hybrid routing methods are more robust since they trade off between computational efficiency and quality of solutions in congestion aware situations. For instance, Hybrid NN2opt achieved a 74.1% optimization rate with a mean planning time of only 7.68 ms, nearly matching the optimal Held-Karp baseline while using 89% less memory. In multi-robot simulations, it reduced collisions by up to 33.33% in congested narrow aisles and up to 100% in wide aisles and maintained more stable performance as robot density increased from 3–5 to 10–15 robots. These results underscore the significance of the inclusion of realistic operation constraints in the modeling of warehouse routing and offer a feasible base on how to enhance the efficiency of order-picking in a big and dynamic warehouse setup.