An Integrated Navigation Approach for Autonomous Driving Vehicles Based on Motion Trajectory Matching*

摘要

For autonomous driving vehicles, smooth and accurate lane-level positioning information is crucial for safe driving. However, it is difficult to achieve high fidelity and smoothness at the same time by using a single navigation approach. Moreover, the state-of-the-art integrated navigation approach mostly exploits Kalman filter, which depends heavily on the accurate positioning results of GNSS. To find a feasible alternative, an integrated navigation approach based on optimal motion trajectory matching is proposed in this paper. The motion trajectory matching is based on an optimization framework, which minimizes the deviation of the raw positioning results between the inertial navigation system and global positioning system. For verification, experiments are carried out in the environment with bad positioning conditions due to tall buildings and woods blocking effects. The results show that smooth and accurate lane-level positioning information could be obtained by applying the proposed approach.