Fuse GPS Course Angle with Quaternion to Improve GPS/IMU-based Velocity Estimation Accuracy

The potential unobservability of the yaw angle in the vehicle velocity estimation based on the low-cost GPS/IMU reduces the estimation accuracy. In contrast, the fusion with the GPS course angle (GCA) can significantly rectify the observability of the yaw angle, thus enhancing the accuracy and robustness of the estimations. Because the GCA contains partial attitude information, it is difficult to directly fuse the GCA with the quaternion, which is a deterministic attitude representation. To solve this problem, the vehicle velocity estimation error state equation based on GPS/IMU is firstly built in the vehicle coordinate system. Furthermore, during the measurement update, the prior estimation of roll and pitch angles and the measured GCA are combined to form a pseudo-attitude, which can be used to realize the fusion of the GCA and the quaternion in the error state-space. The vehicle test results indicate that the fusion of GCA substantially improves the velocity estimation accuracy.