基于视觉信息融合的乒乓球机器人旋球估计及智能击打策略研究

项目名称： 基于视觉信息融合的乒乓球机器人旋球估计及智能击打策略研究

项目编号： No.61305024

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 自动化技术、计算机技术

项目作者： 方灶军

作者单位： 中国科学院自动化研究所

项目金额： 27万元

中文摘要： 基于视觉的打乒乓球机器人是研究高速运动物体的视觉跟踪、轨迹预测和智能拦截策略等关键技术的有效平台。国内外已有多家单位开展相关研究，研制出的原型样机基本实现了非旋转球的简单推挡动作。由于对旋球还缺乏有效的检测与估计方法，机器人击打旋球的成功率还很低。本项目拟充分综合利用现有的视觉信息，重点解决旋球估计和击打问题。本项目的成功实施，不仅可以提高机器人本身的运动竞技能力，也为复杂环境下多信息的综合利用、智能分析和决策、自学习控制等多方面打下理论基础。 本项目的主要研究内容有：(1) 提取挥拍运动轨迹、球拍与球接触时刻的位姿和速度、乒乓球起始段飞行轨迹三方面的视觉信息；(2) 基于视觉测量信息,设计有效的融合算法，估计出旋球的类型和速度大小；(3) 利用旋球估计结果对乒乓球的飞行轨迹和反弹模型进行修正；(4) 设计旋球的回球控制算法，并研究自学习机制，使机器人像人一样逐步适应旋球的击打。

中文关键词： 乒乓球机器人；旋转球；视觉检测；智能击打；轨迹预测

英文摘要： Vision-based table tennis robot is an effective platform to study the visual tracking, trajectory prediction and intelligent interception strategy for the moving object with high speed. At present, several foreign and domestic research communities have carried out the related researches. Most of the prototype robots can strike the ball without spin. Due to the lack of effective measurement and estimation methods for spinning ball, the successful rate of striking the spinning ball is relatively low. This project aims to make full use of the vision information of the spinning ball from several perspectives, and estimate the spinning pattern and the striking strategy. The successful implementation of this project can improve the competitive ability of this robot, and lays the foundation for comprehensive use of the information under complex environment, intelligent analysis and decision, self-learning control, and so on. The main research contents of this project are as follows. (1) The vision information from three perspectives is captured, such as the moving trajectory of the racket, the pose and velocity of the racket when it strikes the ball, the initial flying trajectory of the ball. (2) The spinning pattern and velocity of the ball are estimated based on the fusion of the above three vision information. (3)

英文关键词： table tennis robot；spinning ball；vision detection；intelligent striking；trajectory prediction