自主导购机器人定位与导航

发布时间：2018-04-13 14:19

本文选题：导购机器人 + ROS　；参考：《西南科技大学》2017年硕士论文

【摘要】：随着人类社会的快速发展,机器人技术正逐渐改变着人们的生活方式。针对由商场中商品数量的增多导致购物者很难快速找到欲购买商品的位置的难题,本课题研制了一种基于三轮全向导购机器人的自主定位与导航系统,可以自主地引导购物者到达商场中任意欲购买商品位置。为了使系统具有良好的扩展性及高代码复用率,方便对导购机器人进行后续深入研究,本文选取了目前最通用的ROS(Robot Operating System)作为软件开发平台,对导购机器人关键技术展开研究。首先,针对粒子滤波定位中重采样带来的粒子匮乏问题,本文提出应用粒子群聚散程度来动态添加随机粒子以减小定位误差。同时,通过加入KL距离对粒子集分布进行近似,可以实时地、自适应地改变用于状态估计的粒子数以减轻系统计算量。其次,针对由商场人员较多、环境动态性较大带来的避障难度大的问题,本文把A-star与DWA结合起来应用于导购机器人的路径规划,提高其移动灵活性。同时,针对商场环境复杂性、碰撞风险大的情况,本文将激光扫描和深度图像结合来精确感知环境,不仅具备二维激光扫描仪平面内探测范围广、精度高、实时性好的特点,而且兼有Kinect深度相机三维感知信息丰富的优势。实验证明该方法工作有效,能有效完成三维空间避碰。最后,以三轮全向导购机器人为实验平台,本文实现了前述的定位和路径规划方法。实验结果表明,这些方法可以安全无误地帮助机器人完成自主定位与导航。
[Abstract]:With the rapid development of human society, robot technology is gradually changing people's way of life.In order to solve the problem that shoppers find it difficult to find the location of goods quickly due to the increase in the number of goods in the shopping mall, a kind of autonomous positioning and navigation system based on three-wheeled omnidirectional robot is developed in this paper.You can independently guide shoppers to the shopping mall where you want to buy goods.In order to make the system have good expansibility and high code reuse rate, and facilitate the further study of the purchase guide robot, this paper selects the most common ROS(Robot Operating system as the software development platform, and studies the key technology of the purchase guide robot.Firstly, aiming at the problem of particle scarcity caused by resampling in particle filter localization, this paper proposes a method to dynamically add random particles to reduce the location error by using particle swarm clustering.At the same time, by adding KL distance to approximate the distribution of particle sets, the number of particles used for state estimation can be changed in real time and adaptively to reduce the computational complexity of the system.Secondly, aiming at the difficulty of avoiding obstacles caused by more people in shopping malls and larger dynamic environment, this paper applies A-star and DWA to the path planning of purchase guide robot to improve its mobility.At the same time, aiming at the complex environment and high collision risk, this paper combines laser scanning and depth image to accurately perceive the environment, which not only has the characteristics of wide detection range, high accuracy and good real-time performance, but also has a wide range of detection in the plane of two-dimensional laser scanner.And also has the Kinect depth camera 3D perception information rich advantage.Experiments show that the method is effective and can effectively avoid collision in three-dimensional space.Finally, with the three-wheel omnidirectional robot as the experimental platform, this paper implements the above positioning and path planning methods.Experimental results show that these methods can safely and accurately help the robot to complete autonomous positioning and navigation.
【学位授予单位】：西南科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP242

【相似文献】