New MPC Framework Handles Uncertain CAV Penetration Rates

Conventional dynamic speed limit control for connected automated vehicles (CAVs) assumes the penetration rate is known precisely throughout the control horizon. In reality, observation errors introduce uncertainty in the traffic composition, degrading control performance. To address this, the authors develop a novel model predictive control (MPC) framework that incorporates traffic composition uncertainty into both flow prediction and optimization. They first create an uncertainty-aware macroscopic mixed traffic model where the uncertain penetration rate affects mixed free-flow speed, capacity, and capacity drop conditions via the mixed fundamental diagram. The MPC then optimizes CAV speed limits against multiple admissible penetration rate realizations, improving robustness under heterogeneous conditions.

Simulation experiments on a single-bottleneck freeway corridor and a multi-bottleneck freeway network with merge-diverge interactions demonstrate that the proposed controller generates more spatially coordinated speed limits. This leads to reduced total travel time spent and environmental benefits. The work is published on arXiv (2607.16615) and represents a significant step toward practical deployment of CAV speed limit controls in real-world mixed traffic where the exact CAV share is never fully known.