Based on the slow convergence speed and low coordination mechanism of ant algorithm when it was used to solve robot path planning problem, this paper proposed a new improved ant colony algorithm based on dynamic classification. First of all, it established the population classification model according to the dynamic classification operator. Then, the head-effect strategy with roulette could improve the coordination mechanism and increase the communication between populations. Finally, in order to increase the convergence speed, the dynamic pheromone update strategy with normalization could be taken with the ant algorithm pheromone update formula kept to reflect the influence of elite. In order to verify the effect, it used the grid method to build the robot motion space model. It applied this algorithm to the path planning problem and compared with several intelligence algorithm. The simulation results show that the algorithm can converge quickly and find the optimal path with fewer iterations with relatively high efficiency.

[2019-02-05] Printed Article /doc/2017.08.0716/v2