Performance of a Low-cost Real-time Navigation System using Single-frequency GNSS Measurements Combined with Wheel-tick Data

In the last years, automotive navigation systems combining GNSS measurements with data from other sensors, such as inertial sensors, have become very popular. The measurements of the additional sensors are used to improve the GNSS solution and to perform dead reckoning during periods of GNSS unavailability. We present an embedded solution of Sensor Fusion Dead Reckoning (SFDR) developed for the above-mentioned objectives, using a lowcost single-frequency GPS receiver and wheel-tick (wheelspeed) information available on the vehicle’s bus system. On the basis of wheel-tick measurements from left and right wheel, and using vehicle parameters, the vehicle’s speed and heading rate can be calculated and used for navigation. An in-depth a-priori accuracy analysis on the basis of error propagation was done in order to study the influence of different error sources, to obtain approximate values for the expected dead reckoning accuracies and to specify accuracy requirements for vehicle parameters. We implemented a tightly coupled extended Kalman filter combining GPS and wheel-tick measurements. Wheeltick scaling factors per wheel were introduced as additional unknowns in the state model. The system was designed to enable rear or front wheel-tick navigation. If front wheels are used and the necessary wheel angle information is not available, wheel angles are estimated iteratively. Some of the further challenges to be met in order to achieve highest possible accuracies are timing issues and correct weighting between GPS and wheel-tick measurements. In order to evaluate the performance of the presented SFDR navigation system, numerous road-tests were carried out. Different driving scenarios under varying environmental conditions were investigated, using a high-precision navigation system as reference. We present results of these road-tests and demonstrate performance and limitations of the SFDR system. Compared to the corresponding GPS-only solution, the combined GPS+wheel-ticks sensor fusion solution reaches a significant accuracy improvement, especially for the speed and for position solutions under difficult environmental conditions. The positioning and heading accuracies reached by wheel-tick dead reckoning highly depend on the distance driven and on the heading rate. Under usual driving conditions and starting with error-free heading and position, a heading accuracy of 1-3 degrees and a position accuracy of 1-10 m can be expected after 100 m of dead reckoning.