节点文献
全自主足球机器人的射门与守门算法研究
Reserch on Shooting and Goalkeeping for Autonomous Soccer Robot
【作者】 王伟光;
【作者基本信息】 哈尔滨工业大学 , 计算机科学与技术, 2008, 硕士
【摘要】 足球机器人系统是一类典型的移动机器人的对抗系统,具有实时性要求高、环境高度动态、不可预测性等特点,并将高科技研究和足球联系在一起,通过足球比赛来检验各种理论、方法和技术。本文以全自主足球机器人系统为平台,主要研究了目标的识别与定位和机器人射门与守门算法。视觉系统的目标识别与定位是全自主足球机器人系统的基础,它对机器人感知环境起着关键的作用。HIT-III型全自主机器人的视觉系统是由全维视觉系统和前向单目系统组成的双目异构视觉,两个视觉系统采用基于颜色信息的目标识别,可以独立的对目标进行识别和定位,最后利用卡尔曼滤波进行数据融合。本文把对全自主足球机器人的射门与守门算法的研究转化为对决策系统的研究。在足球机器人系统里,决策系统的任务是将视觉系统传来的场上的数据进行分析,对场上比赛形势做出判断,形成比赛策略,并通过角色的决策模型来规划机器人的运动,完成射门与守门任务。在设计决策系统时,采用了有穷自动机的理论,并构造了各角色的有穷状态机决策模型,每个状态都对应着一个算法,用来控制机器人运动,在状态进行转换时就完成了机器人的运动与相应任务。最后,经射门与守门实验表明,机器人运动轨迹平滑、实时性高,该决策算法可以应用在全自主足球机器人比赛中;并进一步验证了双目异构视觉系统的可行性和经过卡尔曼滤波算法融合后的定位信息的准确性。
【Abstract】 The robot soccer system is a typical competition system of mobile robot, with the character of high request of real time, high dynamic environment, un-predictability, etc., and it associates high technology study with soccer, examining various theories, methods and techniques by the soccer game. This paper is based on autonomous soccer robot system, and the main study is object identification and localization, and shooting arithmetic and goalkeeping arithmetic.Object identification and localization of vision system is the foundation of autonomous soccer robot system, it plays a pivotal role in apperceiving surroundings for robot. The vision system of HIT-III autonomous soccer robot is binocular heterogeneous vision, composed of omni-vision and monocular vision, each of which can identify objects based on color information and localize objects independently, and the localization data is fused by Kalman filter at last.In this paper, studying in shooting arithmetic and goalkeeping arithmetic of autonomous soccer robot is converted into designing a decision-making system. In robot soccer system, the function of the decision-making system is that to analyze the sounding data from vision system, to estimate the state of the match, to make a decision, and to control the robot moving to complete the shooting and defense task. A deterministic finite state automaton is used in designing decision-making system, which is constructed for each role. Every state corresponds with arithmetic, which control the robot to move, so when the state is transferring in deterministic finite state automaton, the robot is working to move and complete task.Finally, experiments showed that, this decision-making arithmetic, with the smooth trajectory of robot motion and high real time, can be used in autonomous soccer robot. And it proves the feasibility of binocular heterogeneous vision and accurately of localization data that is fused by Kalman filter further more.
【Key words】 Autonomous Soccer Robot; Object Identification; Object Localization; Data Fusion; Deterministic Finite State Automaton;