Integrated path controlling and subsidy scheme for mobility and environmental management in automated transportation networks

Abstract

In recent years, as automated vehicle (AV) technologies are advancing at a breathtaking pace, path controlling has engaged increasing attentions among transportation researchers. As an emerging technology for transport management, it employs AVs as mobile actuators to regulate traffic flow on transportation networks, aiming at alleviating inefficiency due to selfish routing. Existing studies have proved the potential of such a path control scheme for mitigating traffic congestion, but research on its competence for environmental management is still lacking. Given above, this paper develops a path control scheme to achieve expected goals of both mobility and emission. Specifically, we assume a fully automated environment where all the vehicles on the network are highly automated and serve as candidates for the path control scheme. We assume that they originally follow the classic user-equilibrium (UE) principle and can be controlled as either system-optimum (SO) or least-emission (LE) users by the central agent. We formulate the mixed traffic equilibrium including the three classes of users as a variational inequality (VI) problem and integrate it as an equilibrium constraint into the optimal control problem. To compensate for the travel utility loss for being controlled, we also propose an integrated path controlling and subsidy scheme, under which travelers who suffer from the control scheme will be compensated with subsidy. The total subsidy, along with the network performance, are considered in the objective function. A sensitivity-analysis-based Frank-Wolfe (SA-FW) method is applied to solve the proposed problems. Using example networks, we test the computational performance of the SA-FW method and examine the control performance of the control scheme under various settings. Based on the numerical results, it is found that considering LE users and environmental benefits renders the control problem more complex than merely achieving SO.