Research Of Multi-robot Path Planning Based On Multi-agent System

In recent years, robots have not only been applied to industry widely, but also stepped in real life gradually. It is changing our life and behavior style. Concerning the core technology of autonomous navigation of intelligent mobile robots, path planning is a key technology connecting users’ applications and robots. Path planning means that robots perceive the environment information by sensors, and completes the user-defined work efficiently by calculating a safety route. In order to overcome the shortcomings of a single robot, such as control, sensors, energy as well as expensive development cost and time cost, to make several simple robots cooperation is an another way to deal with more complex problems. It can also reduce the complexity and increase the reliability. Considering the above aspects, this paper concerns multi-robot path planning, including the following contents:1. Introduce multi-robot research status at home and abroad systematically and describe the existed algorithms about path planning based on their math tool behind. Multi-robot system architecture and path planning methods are also concerned naturally. Further more, the interesting issue on multi-robot cooperative formation control method are discussed and the existed formation control schemes are presented in the end. Both these work lays a solid foundation for the next contents.2. Improve the traditional artificial potential method on path planning to overcome its shortcoming, local minimum, by proposing two comprehensive algorithms. An artificial field method integrated with behavior strategy and An artificial field method combined with improved rapidly-exploring random tree algorithm are used to escape the minimum trap. Moreover, a potential function method is developed to settle the multi-robot collision problem by dividing the multi-robot path planning into two layers. At last, several simulations are presented to show the proposed algorithm practical and reliable using agents with a particle model.3. Research the interesting formation problem in multi-robot path planing using artificial potential field method with virtual leader based on the specific robot, the quad-rotor. The main work include the quad-rotor mathematical modeling, inner loop controller and formation controller design. A simulink model is demonstrating that the several quad-rotors and particle model agents can make a specific formation and avoid collisions and obstacles. At last, the proposed path planing algorithm is realized in the software develop environment to apply to the laboratory quad-rotor platform. Multi-robot experimental system and the related flying tests are added to show the research work.