论文部分内容阅读
针对有人干扰的动态室内环境,利用可定位性估计理论提出一种RGB-D传感器辅助激光传感器的移动机器人可靠自定位方法。利用RGB-D传感器信息快速检测人的位置区域,并通过坐标转换计算激光扫描数据中的动态障碍物影响因子,结合离散化Fisher信息矩阵在线估计观测信息的可定位性矩阵;同时通过预测模型协方差矩阵评价里程计信息的可靠性,从而动态补偿观测信息对粒子集的影响。在典型含多人运动的动态室内环境中实验,结果验证了本文方法能够提高机器人自定位的准确性和可靠性。
Aiming at the dynamic indoor environment disturbed by human beings, this paper proposes a reliable self-positioning method of mobile robot with RGB-D sensor-assisted laser sensor using the theory of locatable estimation. The RGB-D sensor information is used to detect the location area of human being rapidly. The influence factor of dynamic obstacle in laser scanning data is calculated by coordinate transformation. The discretization Fisher information matrix is used to estimate the locatable matrix of observational information online. At the same time, The variance matrix evaluates the reliability of the odometer information to dynamically compensate for the effect of the observed information on the particle set. Experiments in a typical dynamic indoor environment with multiplayer show that the proposed method can improve the accuracy and reliability of robot self-localization.