We propose a novel particle filter-based framework for robust state estimation and integrity monitoring in urban environments using GNSS and odometry measurements. Using a Gaussian mixture model (GMM) for computing particle weights, we develop an expectation-maximization algorithm for jointly inferring GMM weight parameters and a robust particle distribution of receiver state. From the inferred particle distribution and GMM, we determine the navigation system availability based on specified integrity requirements. Unlike traditional residual-based integrity monitoring algorithms that analyze measurement residuals from an estimated receiver position, we incorporate measurement residuals from multiple particles to estimate the receiver position and monitor integrity. Our method achieves small horizontal positioning errors compared to existing filter-based state estimation techniques on challenging simulated and real urban driving scenarios with multiple erroneous measurements. Through multiple simulations, we also show that our method determines system availability with comparable false alarms and missed identifications to best-performing existing integrity monitoring approaches.

翻译：我们提出一个新的粒子过滤框架,用于利用全球导航卫星系统和odo量度测量在城市环境中进行稳健的国家估计和完整监测。我们使用高斯混合模型计算粒子重量,我们开发了一种预期最大化算法,用于联合推算GMM重量参数和接收状态的稳健粒子分布。我们从推断粒子分布和GMM中,根据特定的完整性要求确定导航系统的可用性。与分析估计接收方位置的测量残留物的传统残留完整性监测算法不同,我们采用多种粒子的测量残留物来估计接收方的位置和监测完整性。我们的方法与现有的基于过滤的状态估算技术相比,在以多重错误测量对模拟和实际城市驱动情景提出挑战方面,实现了小规模横向定位错误。我们通过多次模拟,还表明我们的方法决定系统是否具备可比的虚假警报和错漏失识别,以便最佳地执行现有的完整性监测方法。