服务机器人交互式地图构建与路径规划技术研究
发布时间：2018-09-07 13:36
【摘要】：地图构建与路径规划算法是服务机器人自主移动技术的核心;人机交互技术是降低服务机器人使用门槛,普及推广服务机器人的关键。本课题研究了服务机器人交互式地图构建及动态场景下的路径规划技术。目的在于使服务机器人在构建环境地图时,通过语音、视觉等交互方式,实现方便快捷的人机交互;在自主移动时,能有效感知环境中的行人,具备在动态环境中的自主移动能力。本课题的技术可以应用到助老助残、送餐等机器人中,以提高机器人的实用性。在研究过程中,首先针对现有的硬件平台及传感器特性,研究并选择合适的2D栅格地图构建算法与机器人自定位算法;通过实验优化了机器人建图与定位参数。为实现交互式地图构建,课题实现并验证了三种利用Kinect深度相机实现机器人跟随的方案,通过实验从中选出了性能最佳的方案;构建了基于rosjava的移动终端APP,以实现机器人和用户之间的语音交互与建图和路径规划过程的远程监控。为解决服务机器人在动态场景中的路径规划与导航问题,本文在Lattice状态网格搜索空间中,融合行人感知技术获取的行人轨迹预测信息,提高了移动机器人在动态场景中的自主移动能力。课题的研究内容包括服务机器人交互式地图构建技术与基于行人感知技术的移动机器人路径规划系统。先后为机器人配置了光电码盘及9轴IMU以提供精确的里程计信息;采用KCF核化相关滤波算法,实现了鲁棒的机器人跟随行人功能;基于rosjava平台开发了人机交互及机器人远程监控APP;融合激光雷达和深度相机(Kinect)观测信息,实现了鲁棒的行人检测与行人轨迹预测;利用行人感知与轨迹预测信息,构建了动态场景中的机器人路径规划系统。课题构建的服务机器人可以通过人机语音视觉交互APP,实现地图构建与路径规划中的语音标注、机器人跟随、语音控制机器人到达指定位置等功能。地图构建方便快捷,通过APP实时远程监控机器人状态,简洁直观,易于用户操作使用。行人感知模块的行人检测与行人轨迹预测功能,检测迅速,鲁棒性好。机器人具备在动态场景中的自主移动能力,达到了预期的研究目标。
[Abstract]:Map construction and path planning algorithms are the core of autonomous mobile technology for service robots, and human-computer interaction technology is the key to reduce the threshold of using service robots and popularize service robots. In this paper, the interactive map construction of service robot and the path planning technology in dynamic scene are studied. The purpose of this paper is to make the service robot realize the convenient and quick human-computer interaction through the interactive methods of voice and vision when constructing the environmental map, and can effectively perceive the pedestrian in the environment when he moves independently. Ability to move autonomously in a dynamic environment. The technology of this subject can be applied to the robot such as helping the elderly and helping the disabled, feeding and so on, in order to improve the practicability of the robot. In the research process, firstly, according to the existing hardware platform and sensor characteristics, the appropriate 2D raster map construction algorithm and robot self-localization algorithm are studied and selected, and the robot mapping and location parameters are optimized through experiments. In order to realize the interactive map construction, three schemes using Kinect depth camera to follow the robot are implemented and verified, and the best performance scheme is selected through experiments. The mobile terminal APP, based on rosjava is constructed to realize the speech interaction between the robot and the user, and the remote monitoring of the process of map building and path planning. In order to solve the problem of path planning and navigation of service robot in dynamic scene, this paper combines the pedestrian trajectory prediction information obtained by pedestrian perception technology in the Lattice state grid search space. The autonomous mobility of mobile robot in dynamic scene is improved. The research contents include interactive map construction technology of service robot and path planning system of mobile robot based on pedestrian perception technology. In order to provide precise mileage information, the robot has been equipped with photoelectric codec and 9-axis IMU successively, and the robust robot following pedestrian function has been realized by using KCF kernel correlation filtering algorithm. Based on rosjava platform, the human-computer interaction and robot remote monitoring APP; fusion lidar and depth camera (Kinect) observation information are developed to achieve robust pedestrian detection and pedestrian trajectory prediction, using pedestrian perception and trajectory prediction information, The robot path planning system in dynamic scene is constructed. The service robot constructed in this paper can realize the functions of map construction and path planning, robot following, speech control robot to the designated position, and so on through human-computer voice vision interactive APP,. Map construction is convenient, real-time remote monitoring of robot status through APP, simple and intuitive, easy to operate and use. Pedestrian detection and pedestrian trajectory prediction function of pedestrian perception module, rapid detection, good robustness. The robot has the ability to move autonomously in the dynamic scene and achieves the expected research goal.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP242
本文编号：2228416
[Abstract]:Map construction and path planning algorithms are the core of autonomous mobile technology for service robots, and human-computer interaction technology is the key to reduce the threshold of using service robots and popularize service robots. In this paper, the interactive map construction of service robot and the path planning technology in dynamic scene are studied. The purpose of this paper is to make the service robot realize the convenient and quick human-computer interaction through the interactive methods of voice and vision when constructing the environmental map, and can effectively perceive the pedestrian in the environment when he moves independently. Ability to move autonomously in a dynamic environment. The technology of this subject can be applied to the robot such as helping the elderly and helping the disabled, feeding and so on, in order to improve the practicability of the robot. In the research process, firstly, according to the existing hardware platform and sensor characteristics, the appropriate 2D raster map construction algorithm and robot self-localization algorithm are studied and selected, and the robot mapping and location parameters are optimized through experiments. In order to realize the interactive map construction, three schemes using Kinect depth camera to follow the robot are implemented and verified, and the best performance scheme is selected through experiments. The mobile terminal APP, based on rosjava is constructed to realize the speech interaction between the robot and the user, and the remote monitoring of the process of map building and path planning. In order to solve the problem of path planning and navigation of service robot in dynamic scene, this paper combines the pedestrian trajectory prediction information obtained by pedestrian perception technology in the Lattice state grid search space. The autonomous mobility of mobile robot in dynamic scene is improved. The research contents include interactive map construction technology of service robot and path planning system of mobile robot based on pedestrian perception technology. In order to provide precise mileage information, the robot has been equipped with photoelectric codec and 9-axis IMU successively, and the robust robot following pedestrian function has been realized by using KCF kernel correlation filtering algorithm. Based on rosjava platform, the human-computer interaction and robot remote monitoring APP; fusion lidar and depth camera (Kinect) observation information are developed to achieve robust pedestrian detection and pedestrian trajectory prediction, using pedestrian perception and trajectory prediction information, The robot path planning system in dynamic scene is constructed. The service robot constructed in this paper can realize the functions of map construction and path planning, robot following, speech control robot to the designated position, and so on through human-computer voice vision interactive APP,. Map construction is convenient, real-time remote monitoring of robot status through APP, simple and intuitive, easy to operate and use. Pedestrian detection and pedestrian trajectory prediction function of pedestrian perception module, rapid detection, good robustness. The robot has the ability to move autonomously in the dynamic scene and achieves the expected research goal.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP242
【参考文献】
相关期刊论文 前5条
1 张继文;刘莉;陈恳;;面向全方位双足步行跟随的路径规划[J];自动化学报;2016年02期
2 卢先领;王洪斌;徐仙;;基于加速度信号和进化RBF神经网络的人体行为识别[J];模式识别与人工智能;2015年12期
3 张彪;曹其新;何明超;;基于多传感器融合的移动机器人三维地图创建[J];中国科技论文;2013年08期
4 王田苗;陶永;陈阳;;服务机器人技术研究现状与发展趋势[J];中国科学:信息科学;2012年09期
5 宋章军;;服务机器人的研究现状与发展趋势[J];集成技术;2012年03期
相关硕士学位论文 前2条
1 王婷;RoboCup人形组足球机器人的蒙特卡罗自主定位算法研究[D];吉林大学;2012年
2 杨树风;带有机械臂的全方位移动机器人的研制[D];哈尔滨工业大学;2006年
，本文编号：2228416
本文链接：https://www.wllwen.com/kejilunwen/zidonghuakongzhilunwen/2228416.html
