GPS Positioning in the Fast Track: Track Model Constraint Enhancement for OEM4

Sports entertainment requires that instantaneous and continuous measures of the performance of the participants are available to the television audience. In an auto racing environment this includes the relative positions, velocity and diagnostics for the vehicles in the race. Some of the information can be alphanumeric, but a graphical representation gives much more impact to the viewer, and in many cases this is the preferred method of information transfer. This paper discusses a particular application in which position, velocity and vehicle diagnostics are collected from race cars in NASCAR races and transformed so that information can be integrated graphically with the video stream from the cameras at the race and presented to the television viewers when appropriate. Fundamental to the production of graphical information overlays is a set of accurate positions for all the vehicles in the race. In a NASCAR race, there are 43 vehicles, and all of these must be positioned accurately. Accurate relative positions can easily be computed with differential GPS measurements, provided a sufficient number of measurements together giving sufficient geometrical strength are available. Unfortunately, neither of these prerequisites exist at a NASCAR racetrack. So in order to provide continuous and accurate position information, some kind of supplementary measurement is required in addition to GPS. The incorporation of a digital model of the race track into the GPS receiver on every vehicle gives enough additional geometrical strength so that with GPS observations, the accuracy and continuity requirements of the application are satisfied. Once sufficiently accurate position information for all the cars is generated, this has to be transferred to a central location where it is used to generate positions for the vehicles captured in the various video frames. This is transformed to screen co-ordinates so screen graphics can be generated for those vehicles. These graphics are overlaid onto the video images generated by the cameras at the race. In order to do all these tasks in real time, sophisticated telemetry, time synchronization and integration subsystems have had to be designed in addition to the custom GPS receiver software used to generate the vehicle positions in first place. This paper will describe the various subsystems and their integration with particular emphasis on the effect of the track model constraints used to aid the GPS positioning. The methodology for integration of the track model constraints into both the pseudo range and Real Time Kinematic (RTK) filters will be described as well as the method used to transform the “real world” vehicle positions into co-ordinates in the video screen frame. Test results will be provided.