Research On Multiple Mobile Robots Flocking And Coordinative Control

From the intelligent control of mobile robots, the thesis researches the control issues of robot navigation and multiple robots coordination flocking methodically, to indicate the evolvement from robot motion control to multiple robots coordination control, and to solve some particular problems in the existing robotics control methods.At the beginning, the thesis introduces two robot platforms and simulators concerned to the research. From the analysis the three kinds of wheeled mobile robot, the kinematics models of the typical non-holonomic mobile robots are researched. Because the robot platforms: Pioneer II-DXe and Koala are both standard wheeled mobile robots, the thesis designs a trajectory tracking controller based on the error model of the robot's position and attitude, to control the movement of the robot. The tracking effect for a round trajectory is validated in Matlab.Secondly, while the robot's self-determining movement is depended on the sensing of the environment by the sensors, in which the visual signal process play a huge role. While the visual signal process is more complicated and vulnerable, the thesis transforms the color model to reduce the affect from the environment, and then realizes the distance measuring of a moving target based on the three dimensional reconstruction principle. After that, a staged controller is designed to decrease the disturbance of the visual error to achieve the robot dynamic recognition and tracking.Visual signal process is fit for recognizing the object with special color or shape, however, the machine vision has the disadvantage of low precision and procession rate, specially the monocular vision. Oppositely, the laser sensor is widely used for its high precision but easy application. The thesis researches the robot navigation problem based on the laser sensor. First of all, a laser fused visual emendation method is designed to raise the precision of the monocular vision. After that, the laser based robot smooth obstacle avoidance, navigation, and movement plan algorithm are designed based on the laser sensor and are simulated on the Player/Stage platform. By comparing and analyzing the result, the controller can be amended to achieve better efficiency.Furthermore, based on the movement control and sensing decision, the multiple robots coordination is researched. The thesis indicates an extendable multiple robot topologic strategy in the 2D environment, to solve the problem caused by the ambiguous relationship between robots in rigid flocking. Then using the triangular formation for instance, the structure of the formation is defined by a matrix. In this way, the basic mathematic manipulation method can be applied for formation transforming.Topologic strategy needs to be realized by the proper controller, which is the mathematical foundation of multiple robot rigid formation control. Hence, the classic real time pose (position and attitude) feedback based closed-loop controller is adopted, using the local relationship between the robots as the controlled object, and to expand the stable structure to the whole flocking in three ways: cascade interconnection, parallel interconnection and multiple leaders interconnection. A Input-to-State Stability (ISS) theory is researched to analyze the stability of the formation structures above, to achieve the stability condition and ISS stability measure. Meanwhile, to indicate the universality of the strategy, the derivative formations of the standard triangular formation are discussed as well.At last, since the otherness of the robots in different logical position, the rigid flocking strategy is not a distributed strategy essentially. To research multiple robots distributed flocking issue and avoid the drawback of local optimization, the thesis provides an ameliorative artificial potential field method by adding a specific shadow potential field, and indicate a local N-nearest neighbor rule to solve the consensus problem of the multiple robots system on both concentrated flocking and big-scale flocking.