未知环境下机器人语义地图构建

发布时间：2018-03-03 08:44

本文选题：机器人　切入点：未知环境　出处：《东北师范大学》2017年硕士论文　论文类型：学位论文

【摘要】：机器人进入一个未知环境完成智能化任务时,需要感知和熟悉环境,此时就需要对环境进行建模和深层次的理解。这个过程可以通过建立环境地图,并对环境地图进行语义标注,形成语义地图来实现。一个详细、准确、深层次的语义地图是机器人能否快速准确地实现日后智能化任务的前提和关键。构建地图的关键是获取机器人自身的定位信息,但机器人要准确地获取自身定位信息又要基于一个准确的构建好的环境地图。那么如何使机器人在未知自己位置和未构建环境地图时,能够综合利用自身对于位置信息的估计情况以及自身携带的环境感应传感器数据,实现准确的地图构建成为机器人地图研究的重点与难点。本文主要是围绕构建语义地图的具体过程进行研究的,总体上可以分成三部分:未知环境的地图构建、语义地图构建和基于语义地图的任务规划。所做工作如下:综合分析了现有机器人搭载的传感器和一些不可避免的运动误差后,提出并实现了2种基于机器人行位推测的地图构建方法,并利用它们准确快速地构建出未知环境的2维地图。在已建环境地图的基础上,利用机器人机载的语音模块,通过人机语音对话的方式使机器人获取环境中的物品语义信息,建立环境的语义地图,从而解决了机器人感知和熟悉环境并构建语义地图的问题。服务机器人接收用户对它发出的指令,通过语音识别和分词操作,提取出指令中的物品关键词和地点关键词,并应用路径规划算法,在构建好的语义地图上规划出到达目标地点的路径,完成给定的任务。
[Abstract]:When a robot enters an unknown environment to complete an intelligent task, it needs to be aware of and familiar with the environment, and then it needs to model and understand the environment at a deeper level. This process can be done by building an environment map. And carries on the semantic annotation to the environment map, forms the semantic map to realize. A detailed, accurate, The deep semantic map is the premise and key for the robot to realize the intelligent task quickly and accurately, and the key to construct the map is to obtain the localization information of the robot itself. But if the robot wants to get its own location information accurately, it must build a good environment map based on an accurate one, so how to make the robot know its own location and not build the environment map, The ability to synthetically utilize its own estimation of position information as well as the environmental sensor data it carries, The realization of accurate map construction has become the focus and difficulty of robot map research. This paper mainly focuses on the process of constructing semantic map, which can be divided into three parts as a whole: map construction of unknown environment, Semantic map construction and task planning based on semantic map. The work is as follows: after synthesizing the sensors and some inevitable motion errors of the existing robot, In this paper, two methods of map construction based on robot row prediction are proposed and implemented. Using them, 2D map of unknown environment is constructed accurately and quickly. On the basis of built environment map, the airborne speech module of robot is used. The robot acquires the semantic information of the objects in the environment and establishes the semantic map of the environment by the way of man-machine voice dialogue. In order to solve the problem of the robot perceiving and familiarizing the environment and constructing the semantic map, the service robot receives the instruction from the user and extracts the key words of the object and the place from the instruction by voice recognition and word segmentation operation. The path planning algorithm is used to plan the path to the target location on the constructed semantic map to complete the given task.
【学位授予单位】：东北师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP242

【参考文献】