考虑潜变量的自动驾驶汽车租赁行为

将自动驾驶汽车（autonomous vehicle，AV）与分时租赁、点对点（peer-to-peer，P2P）租赁模式相结合，为出行者提供新型的出行方式. 为探究出行者租赁自动驾驶汽车的行为特征，分析出行者对AV分时租赁、AV P2P租赁、私家车、公共交通的选择意愿及其影响因素. 基于出行方式选择意愿的调查数据，将结构方程模型（structural equation model，SEM）与多项Logit （multinomial Logit，MNL）模型相结合，建立同时标定显变量与潜变量参数的结构方程-多项Logit （structural equation-multinomial Logit，SE-MNL）模型，对比分析了MNL与SE-MNL模型的参数标定结果. 研究结果表明:在95%的置信水平下，显变量中的出行费用、车内时间、驾照情况、出行目的、婚姻状况以及潜变量中的便捷性、安全性、乘车体验、舒适性对出行者选择AV分时租赁或P2P租赁的影响都是显著的；SE-MNL模型的拟合度较MNL模型高出2%～3%. 

Leasing Behavior for Autonomous VehiclesConsidering Latent Variables

Combining autonomous vehicle (AV) with time-sharing leasing or peer-to-peer (P2P) leasing, two novel travel modes are provided for travelers. To explore the behavior characteristics of AV rental, travelers’ preferences and influence factors for choosing AV time-sharing leasing, AV P2P leasing, private cars and public transit are analyzed. Based on the preference survey data of travel mode choice, the structural equation-multinomial Logit (SE-MNL) model combining structural equation model (SEM) with multinomial Logit (MNL) model is established, which calibrates the parameters of the observed and latent variables. The comparative analyses are done for the parameter calibration results of the MNL and SE-MNL models. The results show that at 95% confidence level, both the observed variables of travel cost, in-vehicle time, driver’s license status, travel purpose, marital status and the latent variables of convenience, safety, ride experience, ride comfort have significant effects on travelers’ choice of AV time-sharing leasing or P2P leasing. The goodness of fit of the SE-MNL model is 2%～3% higher than that of the MNL model.