Title: Iterative Recommendations based on Monte Carlo Sampling and Trust Estimation in Multi-Stage Vehicular Traffic Routing Games Show Chinese title

Title: 基于蒙特卡洛采样和信任评估的多阶段车辆交通路由博弈中的迭代推荐

Abstract: The shortest-time route recommendations offered by modern navigation systems fuel selfish routing in urban vehicular traffic networks and are therefore one of the main reasons for the growth of congestion. In contrast, intelligent transportation systems (ITS) prefer to steer driver-vehicle systems (DVS) toward system-optimal route recommendations, which are primarily designed to mitigate network congestion. However, due to the misalignment in motives, drivers exhibit a lack of trust in the ITS. This paper models the interaction between a DVS and an ITS as a novel, multi-stage routing game where the DVS exhibits dynamics in its trust towards the recommendations of ITS based on counterfactual and observed game outcomes. Specifically, DVS and ITS are respectively modeled as a travel-time minimizer and network congestion minimizer, each having nonidentical prior beliefs about the network state. A novel approximate algorithm to compute the Bayesian Nash equilibrium, called ROSTER(Recommendation Outcome Sampling with Trust Estimation and Re-evaluation), is proposed based on Monte Carlo sampling with trust belief updating to determine the best response route recommendations of the ITS at each stage of the game. Simulation results demonstrate that the trust prediction error in the proposed algorithm converges to zero with a growing number of multi-stage DVS-ITS interactions and is effectively able to both mitigate congestion and reduce driver travel times when compared to alternative route recommendation strategies. Show Chinese abstract

Abstract: 现代导航系统提供的最短时间路线建议在城市车辆交通网络中助长了自私的路由行为，因此是拥堵增长的主要原因之一。 相反，智能交通系统（ITS）倾向于引导驾驶员-车辆系统（DVS）走向系统最优的路线建议，这些建议主要是为了缓解网络拥堵。 然而，由于动机不一致，驾驶员对ITS缺乏信任。 本文将DVS与ITS之间的互动建模为一种新颖的多阶段路由博弈，其中DVS根据反事实和观察到的博弈结果对其对ITS建议的信任表现出动态变化。 具体而言，DVS和ITS分别被建模为旅行时间最小化者和网络拥堵最小化者，它们对网络状态有非相同的先验信念。 提出了一种新的近似算法来计算贝叶斯纳什均衡，称为ROSTER（基于信任估计和重新评估的推荐结果抽样），该算法基于蒙特卡洛抽样和信任信念更新，以确定博弈每个阶段中ITS的最佳响应路线建议。 仿真结果表明，所提出算法中的信任预测误差随着多阶段DVS-ITS交互次数的增加而收敛到零，并且与替代路线建议策略相比，能够有效缓解拥堵并减少驾驶员的出行时间。