A new time efficient method for fault detection of AVs positioning using PPP RTK

Abstract

Autonomous driving systems require precise real-time positioning capabilities. The estimated positions need to be verified to a certain confidence level where integrity monitoring (IM) plays that significant role. IM mainly detects and removes faults and confirms that whether integrity and continuity requirements are fulfilled or that the estimated position is not guaranteed up to the pre-allocated standards. Although many IM methods are introduced, not all of them are appropriate for real-time applications, particularly when using GNSS phase observations. The main obstacle is that heavy computational load is required in order to consider as fault modes as possible and cover most of threats. This work proposes a new IM process structure that is capable of reducing the computational load significantly compared to other traditional methods. The proposed method is tested on network RTK (NRTK) where PPP-RTK positioning technique is used for both the network and the user. Utilizing PPP-RTK offers: 1) only a uni-directional communications from the network to the user(s) are required, 2) the ability to calculate the undifferenced and uncombined residuals of the observations. The process consists of two stages. First, faulty satellites can be detected, listed and sent to the user(s) taking advantages of the known positions of the network stations. This reduces the risk of experiencing an undetected faults at the user end when applying the second stage where the observations’ residuals are compared to their counterparts of the closet reference station from the network. Observations are ranked among and within satellites based on that ratio. Then exclusion can be attempted based on that ranking. The proposed method was applied on a multi-GNSS data from a low-cost receiver type, which is expected to be used on-baord of vehicles. A promising reduction in the the processing times of up to 98% and 88% compared to Solution Separation and conventional Chi-2 test methods was obtained, respectively.