The article discusses the optimization of the control process of an unmanned vehicle. Currently, there is an active development and use of unmanned vehicles. There is a practice of using unmanned shuttles in closed areas (conferences, forums, etc.). The use of cars with automated control in urban conditions and on rough terrain is being tested. In this regard, it is important to develop control algorithms that allow solving problems of car control in real time under the influence of disturbances and the presence of obstacles. With the development of technology and an increase in computing power, it becomes possible to use optimal control algorithms that allow you to achieve better results when the terminal conditions are met, minimizing energy costs. This paper shows the solution of the problem of optimal control of an unmanned vehicle in the presence of a penalty function, measurement noise and disturbances from incomplete data using the separation principle. The problem of optimal control in a deterministic and stochastic setting is solved using an algorithm with a predictive model with a generalized work functional. The effectiveness of applying the Kalman filter is shown depending on the different intensity of measurement noise and different vehicle speeds. The results of numerical modeling are presented, showing the possibility of using the proposed algorithm to control an unmanned vehicle under various initial and final conditions. The developed algorithm has been successfully applied to bypass a moving object.