Balloon Cars (Soft Cars) used for active-system testing are presently statically following one specific path. Autonomous GPS driven balloon cars can be extremely useful if made to be accurate.Volvo group’s long term plan involves usage of such cars to improve active-safety systems. Thisreport presents comprehensive details about the development of the control system of the soft-car.Software development was preceded by purchase of parts. Detailed descriptions of the ballooncar hardware components are presented. Various hardware components were interconnected toform a CAN Bus network. This thesis work describes CAN networks in general and with referenceto the developed Linux based software system.The navigation system is presented in the thesis report using a ‘bottom up’ approach. First theLow-level functions and variables are described that directly control the hardware component.They are called by high level functions that are subsequently explained. The high level functionsinclude accurate turning and an automated lane change process.The high level functions may be sequentially called and to accurately follow a specific path. Various lines and points may be used in a user defined Cartesian coordinate system. Details about theperformance of various high level functions as well as of a demonstration of a pattern are presented along with the limitations of the overall system.