Predictive Control of Connected Mixed Traffic under Random Communication Constraints

Abstract

Fully connected and automated vehicles have been envisioned to help improve the driving safety and efficiency of the transportation system. However, human-driven vehicles will still be present in the near future, which will lead to connected mixed traffic instead of fully connected and automated traffic. This is challenging because of the complexity of human-driving vehicles and the potential communication constraints in the connectivity. To address this issue, this paper models the connected mixed traffic and proposes model predictive control approaches with various prediction approaches including a new inverse model predictive control (IMPC) based approach to handle random communication delays and packet losses in connectivity. The human-in-the-loop experimental results for connected mixed traffic demonstrated the effectiveness and advantages of the proposed approaches, especially the predictive control with IMPC in handling communication constraints in mixed traffic.