Research And Realization Of Control Algorithm On Wheeled Robot-soccer

The robot soccer is the artificial intelligence and the robot domain extremely rich challenge high-tech crowded project, simultaneously also is an artificial intelligence technology ideal breakthrough point.It combines many aspects such as robotics, artificial intelligence and intelligent control. It is an important experiment platform for studying mufti-agent cooperation and artificial intelligence application technology. The robot plays the soccer, looked resembles the game, actually has demonstrated a national information and the automated technology comprehensive strength.As the auxiliary of developing robot soccer real system, simulation system is paid more attention by people for its economy and flexibility. Because of many issues of difficult to control, high request on real-time, hardware,real robot soccer is very difficult to make the technical movement,it dosen't look like the material competion easily infullenced by the surrounding environment, the offical supplies a unifying platform, therefore, it has a better objectivity.it will benefit to activities competition. Advanced control method is easier inspected and applied on robot soccer simulation competition than real robot soccer competition.The thesis just take the robot soccer competition as background, take the robot soccer simulator as simulation platform, make deep research the motion control system on the robot soccer, and make a series of improvements on the algorithm, the performance and application, and it can act as the base of other simulation and decision-making system.Firstly, the thes??eviewed robot soccer competition, and analyzed the key technologies,present situation research domestic and foreign,scientific research and its meaning on robot soccer. Then,introduced simulation platform and environment, deduced the soccer robot's kinematics model. All that, it provided the model foundation and the platform environment for this article following research.Excellent motion control can make the robot soccer achieve. The conventional PID algorithm has the widespread application in the soccer robot control.The machanism of controling process is complex, precisely set up the kinematic is difficulty,and existing the varying degree the misalignment, time-variable and so on uncertainty,at the same time, along with the request which controls to the robot further enhances, it is very difficult to use the conventional PID control to satisfy system's request.Netural network is a very appealing interdisciplinary, it has strong misalignment mapping ability, parallel processing ability, sef-study ability, so, it obtains the widespread application in the control domain. This article studied and analyzed the BP neural network, BP neural network is the most used neural network at present, and it is a neural network with good performance. Mainly studied the mathematical theoryof BP neural network. Multianalysis some good and bad points of several kind of popular BP neural network study algorithms and some improvement BP algorithms. These research has made the upholstery for the behind research on robot soccer motion control.Then ,we combined this kind of improvement's BP algorithm with PID, and educed a new control algorithm. These algorithm performance distinction were compared, The simulation result indicated, this kind of improvement program compares with several other kind of PID control, it has many advantages,such as small overshoot, quickly adjustable speed ,fast control time, showned that it has the better control performance; in addition, steady state error is small. So, the improvement BP neural network?? has higher control precision, thus, it obtained the satisfied effect.So, applying the BP neural network in PID control can efficiently overcome the limitations of badness of parameter adjusting and poor performance when the plant has nonfinearity, time-varying uncertainty and difficulty in setting up the accurate model.This article studied the structure and algorithm of PID controller based on BP neural network, used improved neural network to PID paramenters to self-adjust on line, designed a stable PID controller which has a ability to self-adjust paramenterPresented the PID controller to the robot soccer to the fixed-point motion and the circular motion. Do many experiments on simulation platform and MATLAB. From the experimental result, we saw that this kind of new PID controller enhanced system's robustness to a certain extent, the robot soccer is stabler and its path is smoother.Path planing has been one of the focuses and difficulties in Soccer robot for long years. In the obstacle condition, seeks for an appropriate path, from the given starting point to destination, the robot soccer can safe and escape from collision bypass all obstacles. Thus, to a great degree, the path planing problem is obstacl-avoiding problem.If the path planing is successful, the robot can swiftly complete the assingned. But if it is failure, robot's motion is blocked, it is difficult to complete action, and it will even influence entire strategy, immediate influence the result of the competition. So,path planing holds the very important status in the soccer robot system.The thesis describes a unique approach of applying a pattern classification technique using support vector machine (SVM) to robot path planning. Support vector machines is a learning machine which is based on the small sample statistical theory.Support vector machines can enhance learning machine's generalization, in addition, it existence glabal optimal solution.We divide obstacles into two classes. SVM generates a non-linear separating surface based on the margin maximization principle.This property is suitable for the purpose of usual path planning problems, that is, generating a safe and smooth path. First, Obtained a group of robot formation, changed them into discrete samples, And then, establishes some sample guidance and the guide spot, the next step, seeks for a feasible way, through the MATLAB simulation, we will obtain one path that can avoid all the obstaclesRegarding different obstacle pattern, the given starting point to destination possibly are in the differe??ntermediate region,therefore when the search step is bigger than some threshold value time, we terminate the search. In the next step search, when could not find the next safety point which fulfils the V < 1conditions, we terminate the search too. After multiple searches, we obtain several path curve curves, By comparing, We will choose the the shortest, the smoothest path curve as the robot soccer's actual path.Simulation experiments show that using SVM could get good effect., the robot can seek a optimal pathAt last, the advancement of our robot soccer system is summarized and indicate achievement significance and insufficiency,then, proposed the forecast present's further work.