In this paper, aiming at the problems that the random mixed working conditions are difficult to identify, the insufficiency of drivability evaluation indicators system and the lack of robustness for the evaluation model, a novel objective evaluation method of drivability for passenger cars is proposed. First, combining the sliding time window method and Naive Bayesian model (NBM), a hybrid working condition recognition model of multi-source sensor information fusion is constructed, and the real vehicle test under random mixed working conditions shows that the accuracy of working condition identification can reach 95.6%. Then, the objective evaluation index for crawling conditions, starting conditions and tip-in conditions are studied according to the suggestions of subjective evaluation engineers, an objective indicator evaluation system considering expert knowledge and information redundancy between indicators is constructed, among them, R-type clustering and principal component analysis (PCA) are used to eliminate the information redundancy between objective indicators. Finally, based on the back propagation (BP) neural network to optimize the fuzzy membership function, an improved fuzzy comprehensive evaluation (Improve-FCE) method considering subjective and objective consistency is designed. Real vehicle test and analysis results demonstrate that the development of intelligent drivability objective evaluation tool (I-DOET) has good reliability and stability, the maximum subjective and objective relative error of the improved FCE model is 5.13%, and the pass rate reaches 93.3%.