Using CP/SMT Solvers for Scheduling and Routing of AGVs
Artikel i vetenskaplig tidskrift, 2021
An improved method for solving conflict-free scheduling and routing of automated guided vehicles is proposed in this article, with promising results. This is achieved by reformulating the mathematical model of the problem, including several improvements and speedup strategies of an existing Benders decomposition method. A new heuristic is also presented that quickly yields high-quality solutions. Moreover, a real-large-scale industrial instance is solved using an open-source satisfiability module theories solver and a commercial constraint programming solver. According to the results, both of these general-purpose solvers can effectively solve the proposed models. Note to Practitioners - The problem of conflict-free routing and scheduling of automated guided vehicles (AGVs) in large-scale manufacturing systems has been an ever-present challenge for many AGV companies. Although these companies have developed rather efficient control policies and algorithms, retrofitting the existing heuristic to future's denser, more complicated, and more demanding AGV layouts is not guaranteed to be easy. Furthermore, the installed system will not necessarily be as efficient as expected. Currently, it is common to use heuristics to allocate vehicles to orders and route them. There are also rules of thumbs to avoid collisions and deadlocks. However, with increasing demand for high-performance AGV solutions, it is of interest to employ optimization algorithms that handle the order allocation, scheduling, and routing in a more efficient way. In this article, we present an improved method to tackle this issue, with promising results. We have developed our work in collaboration with a Swedish AGV company, and we have investigated a real-large-scale industrial instance as our case study.
scheduling
Automated guided vehicle (AGV)
routing
satisfiability modulo theories (SMTs)
constraint programming (CP)