This thesis is created to solve the problem in the transportation field. It is how the vehicle can move from start point to the finish point without causing an accident. Deep Reinforcement Learning with PPO algorithm is used in this thesis. PPO act as an instinct for the agent to choose an action. The instinct will be keep updated until the agent reaches the goal. In this thesis, Unity Engine is used to create the data set or the agent model. Because the problem is included in transportation field, then the data set is in the form of track and the agent model is in the form of a car. Many variables used in this thesis, either for help the analysis process or for updating the instinct of the agent.