With the trajectory modeled in two parts: trajectory curve and acceleration profile, the problems of decisionmaking and trajectory planning for intelligent vehicles are formulated as a non-linear programming(NLP) model to optimize the efficiency, safety, comfort and economy of trajectory. To solve this model, a hybrid intelligent optimization algorithm OODE is developed. With a two-layer framework applied, OODE optimizes the acceleration profile and trajectory curve in the inner and outer layers, respectively. By “roughly” evaluating the candidate trajectory curves, the optimal curve is determined very efficiently. The simulation results show that, the proposed method is capable of handling complicated traffic scenarios with multiple dynamic obstacles, and also can meet the demands of real-time applications. Compared with traditional methods, the model accuracy of the proposed method is higher, and the planning speed is obviously faster.