Vehicle routing optimization problem belongs to NP-hard problem. In view of the low efficiency of the basic ant colony algorithm, the low quality of feasible solutions, and the tendency to fall into local optimal solutions, after fully considering the mathematical model and solution of the general vehicle routing problem, an improved ant colony algorithm with roulette operation and 2-opt optimization operation is proposed. The algorithm performs a second calculation on the probability of the selected path during the operation process, which expands the global search range; at the same time, the obtained path is internally optimized to enhance improved local search capabilities and improve the quality of the solution. The results of simulation experiments through MATLAB software show that compared with the results obtained by the basic ACO algorithm and genetic algorithm, the improved ant colony algorithm has great advantages in performance and solution quality. It can better solve the vehicle routing problem with capacity constraints for the corresponding enterprises to better save logistics costs.