In this article we used simulator experiments to explore the intelligent mechanisms of human decision-making. Three types of typical emergency scenarios were used in the experiment, in which Scenario 1 was used to analyze the driver’s choice to protect themselves or to protect pedestrians in emergency situations. Scenario 2 was compared with Scenario 1 to verify whether the driver’s avoidance behavior to protect pedestrians was instinctive or selective. Scenario 3 was to verify whether the driver would follow the principle of damage minimization. The driver’s decisions and actions in emergency situations, from the cumulative frequency of time to collision (TTC) to the maximum steering wheel angle rate during the experiments, were recorded. The results show that the driver was not just instinctively avoiding the immediate obstacle, but more selectively protecting pedestrians. At the same time, the time taken up by the driver’s instinctive avoidance response also had a negative impact on decision-making. The actual decisions of the driver were analyzed to provide a basis for building up the ethical decision-making of autonomous vehicles.

Objective：The moral and ethical issue is a great challenge to the development of autonomous vehicles. There may be distinctions between the choices made by an observer and a participant. The paper is designed to investigate whether drivers will sacrifice the fewest people to save more people in social dilemma, and whether human drivers would give priority to protecting pedestrians or self-protection in an emergency. Methodology: The experiment was conducted with a total of 50 participants assigned to three groups. Three experimental scenarios were designed and each of them contained a social dilemma. A driving simulator was used in this study to explore the choices of human drivers in social dilemma. In addition, the simulator results were compared with those of questionnaire survey. Result: In study 1, 73% of 22 participants swerved into the right lane to hit only one pedestrian for the safety of other five. In study 2 and 3, more participants chose to hit the barrier to protect the pedestrian. Conclusion: A conclusion can be drawn from the second and third group of experiments that most drivers consider not only their own safety, but the safety of pedestrians. Most of the participants intended to minimize the total amount of harm in social dilemma. The choice of crashing into barriers to protect a pedestrian can also be seen as a way to minimize the total amount of harm.

We present a novel pedestrian count estimation approach based on global image descriptors formed from multi-scale texture features that considers spatial distribution. For regions of interest, local texture features are represented based on histograms of multi-scale block local binary pattern, which jointly constitute the feature vector of the whole image. Therefore, to achieve an effective estimation of pedestrian count, principal component analysis is used to reduce the dimension of the global representation features, and a fitting model between image global features and pedestrian count is constructed via support vector regression. The experimental result shows that the proposed method exhibits high accuracy on pedestrian count estimation and can be applied well in the real world.