Autonomous navigation of vehicle systems is treated as a problem of designing a proper energy landscape. By means of an analogy of streamlines in fluid dynamics, the energy landscape for general autonomous navigation is represented in the form of vector fields. We first show that the energy landscape derived from a two dimensional doublet in fluid dynamics provides a natural design for the controller in the truck back-upper problem in free space. By extending this crucial design, we then obtain a systematic method for designing the emergent flow fields (EFFs), which essentially depict an energy landscape satisfying the requirements of autonomous navigation. Information within a navigation environment, including the final state of the vehicle system, multiple irregular obstacles and boundary conditions, are naturally contained in the distribution of the obtained energy landscape. While taking the kinematic constraints of the vehicle system into account, a gradient trace along the direction of our emergent flow field reaches the global minimum of the energy landscape, which corresponds to the final state of the required navigation. Furthermore, the obtained energy landscape has no local minima. Computational efficiency in implementing the EFF approach and direct use of raw images make it a viable method for autonomous navigation of a vehicle system within a complex environment in real time.