An improved ant colony algorithm is proposed for capacitated vehicle routing problems(CVRPs). A new
initialization of vehicle’s position with an optimal and random selection increases the possibility of obtaining the optimal
path. In the process of searching, the ants are more sensitive to the optimal path, because the saving path among customers is
chosen as the heuristic information. The method of local and global dynamic phenomenon update is used in order to adjust
the distribution of phenomenon according to vehicle routes. Except the method of 2-opt, insertion and exchange search
methods are also used to expand the scope of the search for the clients on different vehicle visits. The simulation results
show the effectiveness of the proposed algorithm.