Abstract:The advent of automated vehicles (AVs) and connected automated vehicles (CAVs) creates significant uncertainties in infrastructure planning due to many unknowns, such as performance variability and user adaptation. As technologies are still emerging with low market penetration, limited observational data hinder validation and escalate prediction uncertainty. This study addresses these gaps by employing diverse vehicle models and wide performance ranges in Aimsun microsimulations. It involved three AV/CAV car‐following models with the default Gipps human‐driven vehicle (HDV) model. We evaluated the performance of a mixed fleet in three well‐calibrated real‐world corridor models, including two highways and one freeway. Vehicle parameters in Aimsun are commonly drawn from a corresponding truncated normal distribution with fixed mean, min, and max values. However, to account for future uncertainty and heterogeneity, our AV/CAV models were given truncated normal distributions with variable means for important parameters to incorporate broader performance ranges. The variable means are drawn from intervals with uniform probability, and some of the interval extended below HDV values to account for scenarios where riders opt for smoother rides at the cost of traffic flow. Recognizing that precise future prediction is unattainable, we aimed to establish traffic performance boundaries that define best‐ and worst‐case scenarios in a mixed‐fleet environment. Enumerating all possible combinations is impractical, so a refined optimization algorithm was employed to expedite solution discovery. Our findings suggest that AVs/CAVs, even with conservative performance parameters, can improve traffic operations by reducing peak delays and enhancing travel time reliability. Freeways benefited more than arterial roads, especially with full CAV penetration, although the authors speculate this could create bottlenecks at off‐ramps. The added capacity may induce traffic demand that is difficult to estimate. Instead, we conducted a demand sensitivity analysis to gauge additional traffic accommodation without worsening delays. Compared to point predictions, establishing the range of possibilities can help us future‐proof infrastructure by considering uncertainties in the planning process. Our framework can be adopted to test alternative models or scenarios as more data becomes available.

Bayesian Methods in Automated Vehicle's Car-following Uncertainties: Enabling Strategic Decision Making

A strategic approach to handle performance uncertainties in autonomous vehicle's car-following behavior

Real-time Monitoring of Autonomous Vehicle's Time Gap Variations: A Bayesian Framework

Adaptive robust control algorithm for enhanced path-tracking performance of automated driving in critical scenarios

Integrated Decision Making and Trajectory Planning for Autonomous Driving Under Multimodal Uncertainties: A Bayesian Game Approach

Risk-informed decision-making and control strategies for autonomous vehicles in emergency situations

Combining Belief Function Theory and Stochastic Model Predictive Control for Multi-Modal Uncertainty in Autonomous Driving

Understanding heterogeneity of automated vehicles and its traffic-level impact: A stochastic behavioral perspective

Modeling Uncertainties for Automated and Connected Vehicles in Mixed Traffic

Incremental Bayesian Learning for Fail-Operational Control in Autonomous Driving

Unravelling uncertainty in trajectory prediction using a non-parametric approach

Prediction-Uncertainty-Aware Decision-Making for Autonomous Vehicles

Uncertainty-Aware Prediction and Application in Planning for Autonomous Driving: Definitions, Methods, and Comparison

Uncertainty-Aware Driver Trajectory Prediction at Urban Intersections

On the Robotic Uncertainty of Fully Autonomous Traffic

Uncertainty-Aware Decision-Making for Autonomous Driving at Uncontrolled Intersections

Uncertainties in Onboard Algorithms for Autonomous Vehicles: Challenges, Mitigation, and Perspectives.

Aware of Scene Vehicles—Probabilistic Modeling of Car-Following Behaviors in Real-World Traffic

Assessing the Safety and Reliability of Autonomous Vehicles from Road Testing

Probabilistic Situation Assessment for Intelligent Vehicles with Uncertain Trajectory Distribution

Spatial Uncertainty Sampling for End-to-End Control