为实现低成本高精度的室内定位,在综合考虑定位需求和计算复杂性的基础上,设计了一种基于简易视觉标签的室内高精度定位技术。该技术采用简单的颜色和形状特征检测视觉标签,从而降低检测复杂度,减少数据存储需求。进一步针对简易标签特征不唯一问题,设计了基于相机视场和标签方位角的快速查询匹配方法。并通过分析标签分布特性与定位误差关系,设计加权最小二乘位姿估计算法,建立迭代求解和最优估计的协同策略,实现了算法复杂度和定位精度的良好平衡。仿真和实验结果表明所提定位方法能够有效处理超定方程组奇异问题,抑制不良标签组的负面影响,与超宽带技术相比定位误差降低超62%。

To achieve low-cost and high-precision, an indoor location system with simple visual labels is designed by comprehensively considering the requirement and computation complexity. Only the color and shape features are used for label detection, by which both detection complexity and data storage requirement are reduced. To deal with the nonunique solution problem caused by the simple label features, a fast query and matching method is further presented by using the view field of camera and azimuth of label. Then based on the relationship analysis between the distribution characteristics of labels and location error, a pose and position estimation method using the weighted least square algorithm is designed and cooperated with the interactive algorithm by the designed switching strategy. A good balance has been made between the algorithm complexity and location accuracy by the presented techniques. The simulation and experiment results show that the proposed method can deal with the singular problem of the overdetermined equations effectively and attenuate the negative effect of bad label groups. Compared with the ultra-wide band technology, the location error is reduced more than 62%.

国家重点实验室开放基金(KFY2209)；汽车协同创新中心揭榜挂帅项目(2022CDJDX-004)和重庆市技术创新与应用发展专项( CSTB2022TIAD-KPX0139)