Figure 3.Schematic diagram of geometric relation of PP
2.3 Stanley’s algorithm
The principle of Stanley algorithm is to control the front wheel turning angle φ according to the lateral position error ye and heading angle error from the front axle center control point to the nearest reference trajectory point (gx, gy), and Figure 4 shows the geometric relationship diagram of Stanley algorithm.Stanley algorithm has good low-speed driving stability and small path tracking error, but there is the problem of not being able to meet the path tracking accuracy and smooth line requirements at the same time. problem, and in order to further improve the control performance of the algorithm, the Stanley algorithm is usually improved [36].
For example, in the literature [37], the Stanley controller was optimized by genetic algorithm; Ahmed AbdElmoniem et al. proposed to add a predictive Stanley controller that mimics the driver’s behavior under low-speed operating conditions [38]; Sun et al. proposed an improved fuzzy Stanley model based on particle swarm optimization for unmanned operation of agricultural machines [39]; Wang et al. designed an improved Stanley algorithm [40]. These improved Stanley controllers are usually able to reduce lateral errors and improve yaw stability compared to the original Stanley controller, resulting in a large improvement in control performance.
The Stanley algorithm requires a higher degree of path smoothing and is prone to the problem of excessive vehicle response overshoot in the case of unsatisfactory road curvature smoothness, and the tracking performance is poor when the vehicle has high lateral acceleration due to the neglect of vehicle dynamics and steering actuator dynamic characteristics.