The work aims to enable the glowworm swarm optimization (GSO) algorithm to be applied to the solution to the vehicle routing problem (VRP) and improve the solution performance of GSO algorithm. Based on the improvement of GSO algorithm, the chaotic simulated annealing GSO (CSAGSO) algorithm was put forward to solve the VRP. Firstly, the improved GSO (IGSO) algorithm which enabled the IGSO algorithm to adapt to the solution to VRP was designed; secondly, the simulated annealing mechanism was introduced into the IGSO algorithm, and the simulated annealing GSO (SAGSO) algorithm was proposed, which made the local optimal solution of IGSO algorithm jump out of local optimum. Then, the chaotic mechanism was introduced into the SAGSO algorithm, and the CSAGSO algorithm was proposed, which carried out the chaos initialization and chaos perturbation of the fluorescein concentration value of the SAGSO algorithm. Finally, simulation tests were carried out on a standard example set. Compared with genetic algorithm, ant colony algorithm and particle swarm optimization algorithm, the global optimization ability, convergence rate and stability of CSAGSO algorithm were improved by more than 50%. The improvement of GSO algorithm is reasonable, and the global optimization ability, convergence rate and stability of CSAGSO algorithm are better than those of the genetic algorithm, ant colony algorithm and particle swarm optimization algorithm.