There are two contradictory issues in driving decision-making models: over-simplification hypothesis on adjacent driving zone, and over-complication mathematical modeling on car-following and lane changing. This paper aims to make improvements on these two issues. With the assumption that drivers tend to seek higher speed and more space simultaneously, this paper proposed a comprehensive microscopic driving decision-making model, which integrated inert driving, car-following, and lane changing models. Different from traditional driving decision model based on vehicle trajectory, the new model defined a new physics term spatial motion energy. With this definition, a driving decision function was derived from the principle of minimum potential energy. The model reproduced the decision-making process, in which a discretionary driving vehicle can voluntarily switch between preemptive car-following, lane changing, and non-preemptive car-following (inert driving) tasks. The extended decision-making models introduced new parameters and boundary conditions, which can solve the problems of driver heterogeneity and mandatory driving mode decision. Compared with the traditional driving decision models, the novel model is more consistent with the real road traffic scenarios, and the decision function is more comprehensive. Moreover, a ternary tree framework was developed to facilitate the development of computationally efficient self-driving programs. This framework may promote the applications of the novel model on autonomous driving, especially for the driving decisions of autonomous vehicles above Level 3 in multi-lane highway driving scenarios. Furthermore, insurance companies and actuaries can use this new model to evaluate the safety of the driving behavior of autonomous vehicles or traditional cars driven by human drivers, effectively distinguishing aggressive driving style from conservative driving style.