Impact of Positioning Uncertainty on Eco-Approach and Departure of Connected and Automated Vehicles
Nigel Williams, Guoyuan Wu, University of California - Riverside; Pau Closas, Northeastern University
Alternate Number 1
Connected and Automated Vehicles (CAVs) have been well regarded as enabling technologies to solve many existing issues in contemporary transportation systems, including minimization of accidents, mitigation of congestion, reduction of energy consumption and pollutant emissions, as well as enhancement of system resilience. Numerous studies have been focused on the development and evaluation of various CAV applications, either based on field testing within a limited scope or using traffic simulation at a relatively large scale. The vast majority of such applications require positioning technologies to provide accurate and reliable CAV position estimates. This information is typically provided by Global Navigation Satellite Systems (GNSS) in conjunction with other sensors (typically Inertial Measurement Units, or IMUs for short) and contextual information (for instance, map matching). It is agreed that GNSS is an enabler of CAV applications, where positioning is a fundamental component in many use cases. However, many of such applications are conceived assuming perfect knowledge of the vehicle’s position, omitting that GNSS positioning is not error-free. These errors are on the order of several meters and could go down to decimeters, when real-time kinematics (RTK) or integration with IMU is considered. To our best knowledge, 1) most CAV applications, especially environment-focused ones, rely on the assumption that accurate positioning is continuously available. Nevertheless, this is not the case in real-life situations, where the accuracy of vehicles’ positions is estimated. For instance, if the vehicle computes its position through GNSS, this may be degraded in urban canyons due to blockage of GNSS signals and multipath effect; and 2) very little research has been conducted to evaluate the impacts of positioning accuracy on the development of CAV applications at the traffic level. As mentioned, many CAV applications are designed under the strong assumption that GNSS positioning is errorless.
To address these research gaps, we used a state-of-the-art environmentally friendly CAV application, called Eco-Approach and Departure (EAD) at Signalized Intersections, as a study case. The basic idea of the EAD application is to take full advantage of upcoming traffic signals’ phase and timing information available via infrastructure to vehicle (I2V) communication, and then provide the equipped vehicle with longitudinal speed guidance to pass through the signalized intersection in an environmentally friendly manner (e.g., to avoid unnecessary stop-and-go and idling maneuvers). The goal of this article is to evaluate the effectiveness of EAD at the traffic level under positioning uncertainties and to get more in-depth understanding on how positioning accuracy would affect the system performance of CAV applications. To achieve the objective, we first characterized GNSS-based positioning errors (both longitudinally and laterally) and introduced the associated disturbances to those key system parameters, for instance, distance to the stop location and index of lane where the equipped vehicle was traveling. Then, we performed impact evaluation in PARAMICS, a microscopic traffic simulation environment, where a three-intersection signalized corridor was coded against a real-world testbed along El Camino Real in Palo Alto, California, and the traffic pattern and signal control were well calibrated with field data. The EAD application and position disturbance mechanism were implemented via an Application Program Interface (API). We examined the system effects in terms of safety (e.g., number of potential conflicts), mobility (e.g., average trip delay) and environmental sustainability (i.e., energy consumption and pollutant emissions), under different levels of positioning accuracy, traffic volumes, and penetration rates of EAD application.
Our findings indicate that the EAD metrics can be moderately degraded due to inaccurate position estimation at the vehicles, both in terms of energy consumption and travel time. On the one hand, this should stimulate further research on more accurate and reliable positioning solutions to enable the CAV community. On the other hand, it highlights the necessity to account for such position uncertainty when designing CAV applications.