Study On Multi-robot Path Planning Based On An Improved Ant Colony Algorithm

Robot path planning means that every robot plans out a safe and no collision optimal path in its obstacle environment according to certain evaluation criteria. And it is an important branch of robotics and it is the precondition and foundation of perform tasks. multi-robot system have big advantages of accomplishing complex task cooperatively over single robot. Currently, most of the path planning research is still concentrated in single robot planning in a static environment. Path planning in dynamic and multi-robot system environment are still big questions for further study.This paper aims to solve the issue of path planning in multi-robot system and dynamic environment. The global planning of single robot technology and obstacle collision avoidance methods of single and multi-robot in dynamic environments are studied. A method based on dynamic priority rules, an improved ant colony algorithm and rolling window bilevel programming to solve the multi- robot system path planning is been proposed. The main works are as follows:(1) By the analysis of the basic ant colony algorithm, one kind of improved ant colony algorithm based on the basic ant colony algorithm is proposed to make it better adapt to the global path planning of the robot. The improvements include adaptive heuristic function, the state selection strategy and pheromone distribution mechanism. Several simulation experiments based on improved ant colony algorithm in one static environment is conducted to show the superiority of the proposed algorithm.(2) In order to solve the problems of robot obstacle avoidance planning in dynamic environment, one method combining rolling window with bilevel programming is proposed. In the first layer, global path planning approach, an improved ant colony algorithm, is applied to plan global path in a static environment. In the second layer, local collision avoidance methods are used to avoid obstacle in rolling window. The methods of forecasting dynamic obstacles, collision avoidance and emergency robot collision avoidance strategy are developed. At last single robot dynamic collision avoidance simulation experiment is accomplished in the simulation software. And the experimental results demonstrate the effectiveness of the proposed method.(3) A dynamic priority rule, which is made according to the robot remaining path length of every robot, is proposed to avoid paths conflicts in multi-robot system. Obstacle prediction method and the robot collision avoidance strategy in multi-robot system are developed. The dynamic priority rules and rolling window and bilevel programming algorithm are combined to solve multi-robot system path planning problem. Finally, a simulation environment is designed, and the simulation results show the effectiveness of the method.