摘要
为了解决未知环境下的单目视觉移动机器人目标跟踪问题,提出了一种将目标状态估计与机器人可观性控制相结合的机器人同时定位、地图构建与目标跟踪方法。在状态估计方面,以机器人单目视觉同时定位与地图构建为基础,设计了扩展式卡尔曼滤波框架下的目标跟踪算法;在机器人可观性控制方面,设计了基于目标协方差阵更新最大化的优化控制方法。该方法能够实现机器人在单目视觉条件下对自身状态、环境状态、目标状态的同步估计以及目标跟随。仿真和原型样机实验验证了目标状态估计和机器人控制之间的耦合关系,证明了方法的准确性和有效性,结果表明:机器人将产生螺旋状机动运动轨迹,同时,目标跟踪和机器人定位精度与机器人机动能力成正比例关系。
To address the object tracking issue of a mobile robot based on monocular vision in an unknown environment,a method for robotic simultaneous localization,map building,and object tracking is proposed,combining the object state estimation with the observability control of a mobile robot.Regarding state estimation,based on robot monocular visual simultaneous localization and mapping,a target tracking algorithm under the extended Kalman filtering framework is designed.Considering the observability control,an optimal control method based on updating the maximization of the target covariance matrix is designed.Using the monocular vision,this method can synchronously estimate the robot’s own state,environmental state,and target state,as well as the target following.The simulation and prototype experiments verify the coupling relationship between the target state estimation and robot control,demonstrating the effectiveness and accuracy of the method.The results indicate that by employing this method,the robot can generate a spiral maneuvering trajectory,and the accuracy of target tracking and robot positioning is directly proportional to its maneuvering ability.
作者
伍明
李广宇
魏振华
汪洪桥
WU Ming;LI Guangyu;WEI Zhenhua;WANG Hongqiao(Information System Office,Xi’an High Technology Research Institute,Xi’an 710025,China;Teaching Department,Shaanxi Radio and TV University,Xi’an 710023,China)
出处
《智能系统学报》
CSCD
北大核心
2022年第5期919-930,共12页
CAAI Transactions on Intelligent Systems
基金
国家自然科学基金项目(61503389)
陕西省自然科学基金项目(2015JM6313,2020JM358)。
