A lifelong framework for data quality monitoring of roadside sensors in cooperative vehicle-infrastructure systems

Abstract

To monitor the data quality of roadside sensors in cooperative vehicle-infrastructure systems (CVIS), this study proposes a lifelong framework based on high-precision positioning and perception data of fully instrumented connected and automated vehicles (CAVs). First, a novel trajectory similarity algorithm, called longest common subsequence considering time and relative position sequences (LCSS-TRPS), is developed to match the CAV perception data with roadside perception data. The system time deviation is then calculated, and Kalman filtering is applied to synchronize the sampling time. Finally, indicators are rigorously designed considering absolute and relative positioning errors to assess the data accuracy. Simulation via PreScan and field experiments on Donghai Bridge (China) are conducted to verify the performance and feasibility of the proposed framework. The results show that the algorithms of trajectory matching and time synchronization are efficient and stable under different conditions, and the accuracy of data can be effectively evaluated by the designed indicators.