Abstract:

The yaw rate measured by on-board angular velocity sensor is inevitably contaminated by sensor noise， and is also with hysteresis. In order to improve the accuracy of yaw rate estimation， this paper presents an estimation algorithm based on Reliability Indexed Sensor Fusion （RISF） multi-source sensor information fusion . Firstly， the yaw rate sensor’s measured value is filtered by an algorithm of adaptive cubature Kalman filter（ACKF）. Secondly， the yaw rate is estimated using kinematics method by a single track model， which takes road bank angel into account. Through this model， recursive formulas are established using velocity， front wheel steering angel and lateral acceleration as input， and also using Ackermann steering geometry's output as update value. Lastly， an adaptive Kalman filter based on RISF （RISF-AKF） is applied to fuse the filtered value with the model estimation. A real vehicle road test shows that the RISF-AKF method can estimate road bank angel precisely， and the RISF fusion value has a better performance than the single sensor processed value.

Key words: adaptive cubature Kalman filter, single track model, Ackermann steering geometry, RISF, yaw rate estimation