【摘要】 针对无人农业机器人在复杂作业环境中因频繁非视距（non line of sight,NLOS）通信导致超宽带（ultrawide band,UWB）定位系统量测波动大、精度低的问题，提出一种改进误差状态卡尔曼滤波（error-state Kalman filter,ESKF）的UWB与惯性导航单元（inertial measurement unit,IMU）紧耦合定位技术。首先，采用非对称双面双向测距法结合线性拟合校准优化UWB量测数据，设计基于改进的均值滤波算法剔除离群值；其次，基于改进ESKF框架实现UWB-IMU协同定位，利用IMU状态预测信息构建自适应因子，动态调整量测噪声协方差矩阵以削弱NLOS误差影响；最后，搭建四轮无人农业机器人平台，在典型NLOS农业场景下进行静态及动态目标定位试验验证。结果表明，在动态轨迹跟踪中，相较于纯UWB和传统EKF算法，总体定位精度分别提升53.38%和25.15%。该方法在复杂遮挡环境下具有良好的鲁棒性，可为无人农业机器人实现高精度自主导航定位提供技术支撑。更多还原

【Abstract】 To address the issue of large measurement fluctuations and low accuracy in ultra-wide band(UWB) positioning systems of unmanned agricultural robots operating in complex environments due to frequent non line of sight(NLOS) communication, an improved error-state Kalman filter(ESKF) tightly coupled UWB and inertial measurement unit(IMU) positioning technique was proposed. Firstly, an asymmetric bidirectional ranging method combined with linear fitting calibration was used to optimize UWB measurement data, and an improved mean filtering algorithm was designed to remove outliers. Secondly, UWB-IMU collaborative positioning was achieved based on the improved ESKF framework. An adaptive factor was constructed using IMU state prediction information to dynamically adjust the measurement noise covariance matrix to mitigate the impact of NLOS errors. Finally, a fourwheeled unmanned agricultural robot platform was built, and static and dynamic target positioning experiments were conducted in typical NLOS agricultural scenarios to verify the technology. The results showed that, in dynamic trajectory tracking, the overall positioning accuracy was improved by 53.38% and 25.15% compared to pure UWB and traditional EKF algorithms, respectively. This method exhibited good robustness in complex occlusion environments and could provide technical support for high-precision autonomous navigation and positioning of unmanned agricultural robots.更多还原