The automotive market today has seen the entry of Level-3 conditional automated driving vehicles equipped with an automated driving system that waits for the drivers to start it on the road. Before making a full assessment of the use of automated driving systems, drivers should be made to experience real-world conditional automated driving. A driver may have a mood change when driving a real-world automated vehicle. This emotion points to the mediation of motivation, which affects a driver’s cognition and intention to start an automated driving system on the road. In this study, the emotion of experiencing autonomous driving, cognition, and satisfaction of the driving performance were introduced to construct an intention model to start an automated driving system. Online and off-line questionnaires were adopted, and the emotional response, cognition of automated driving, and intention of 133 drivers who experienced real-world conditional automated driving were determined. Driver experience was assessed in four scenarios as part of emotional tests: during manual driving, during conditional automated driving, during takeover under the influence of the warning system, and during takeover driving. The results of the questionnaire showed a significant positive correlation between emotion and cognition, satisfaction of autonomous driving performance, and the intent to start the automated driving system. Emotions play a mediating role between cognition, satisfaction, and intention to start automated driving. Drivers who experienced conditional automated driving appeared to exhibit a moderately high level of emotional response in terms of joy, interest, and surprise, whereas medium-level negative emotions included fear and anger. Drivers experienced some intensity of emotional changes during conditional automated driving and takeover driving. The emotional changes were uneven but encouraging support was reported. In addition, specific hypotheses relating the driving performance of the automated vehicles (in terms of programmed design of takeover and warning system of takeover) to the emotional dimensions were tested. A cluster analysis of the emotional response measures revealed five different emotional patterns when experiencing the real-world automated vehicle, among which the happy/satisfied group had higher intention to start an automated driving system on the road, followed by the emotional group, whereas the disgust group showed the lowest intention. The cluster analysis was supported by demographic and driving cognitive characteristics (age, education, and self-evaluation of the driving level and driving experience) of the five groups of drivers. Finally, the theoretical and practical significance of this study was expounded. The research results may provide some suggestions and hints for the government and enterprises to promote the development of automated driving.