Tridimensional vector path abstracting and trajectory tracking control on ramps of full self-driving vehicle

Abstract

Under the constraints of time and the corresponding vehicle location, trajectory tracking control of full self-driving vehicles on ramps requires higher speed to overcome the increase in the driving distance caused by the road slope. Two-dimensional trajectory tracking methods are not applicable to roads with various gradients in reality. Here we discuss a series of studies on tridimensional vector path abstracting and trajectory tracking, collectively designing a network physical model with meaningful nodes for describing real roads and a trajectory tracking controller suitable for ramps. Almost all kinds of paths can be characterized by several simple polynomials, which are then used as the path demand in trajectory tracking control. A vehicle kinematics model considering its vertical location and slope angle attitude is formulated, based on which, model predictive control algorithm is used for the trajectory tracking controller design by regulating the vehicle speed and front wheel steering angle. The developed approach is downloaded into an intelligent control unit and tested in the real world by using a self-driving test vehicle to fully realize the practical application.