In this paper, we propose a constrained optimal control approach as a reference trajectory generator for a driving scenario with uncertainty. Given a scenario, this generator can produce a reference trajectory in order to make validations for autonomous vehicles decision making problems. The constrained optimal control problem guarantees to obtain a collision-free trajectory with safety and comfort based on the design of the objective function and the constraints of the vehicle. The uncertainty of environmental information provided by sensors is taken into account, and a stochastic optimization problem is proposed to limit the risk of violating safety requirements. Numerical experiments show that the stochastic model can better ensure the robustness of the obtained solutions.