Route choice behaviour and lane management strategy considering capacity variability for mixed flow of autonomous vehicles

The rapid development of autonomous vehicles (AVs) has introduced mixed traffic flows of AVs and human-driven vehicles (HVs) into urban transportation systems. While existing network-level traffic assignment studies emphasise the enhancement of road capacity due to AVs, the impact of capacity variability has been relatively underexplored. This study addresses this gap by introducing capacity variability and tracing its influence on traveller route choices and lane management strategies. We extend the probabilistic user equilibrium (PUE) model to accommodate the mixed AV-HV traffic flow. A bi-level optimisation framework is developed to determine the optimal lane allocation. Our results demonstrate that: (1) capacity variability significantly affects both travellers’ long-term route choices and optimal lane allocation strategies, and (2) AVs reduce system-wide travel costs and the sensitivity of system performance to capacity variability. These findings offer important policy implications for urban mobility planning in the era of AV adoption.