Study Onadaptive Vehicle Speed Control Based On Road Geometric Characteristics

Recently, with the rapid mobilization in China, road safety becomes increasinglycontentious. Increasing number of vehicles and complex geometric characteristics ofroads, such as steep slope, sharp curve and irregular road surface, result in a largenumber of traffic accidents. As a key component of Intelligent TransportationSystems (ITS), intelligent vehicle covers a wide range of functions of active safety,such as advanced driver assistance and autonomous driving, which can increase roadcapacity, improve safety, and reduce energy consumption and pollution. The adaptivevehicle speed control based on road geometry has been one of key challenges ofintelligent vehicle technologies. Hardware-in-the-loop simulation (HILS) hasunparalleled advantages in convenience, reproducibility, adaptability and safety of thesystem design and test over vehicle experiments in the field. In order to tackle thischallenge, this dissertation presents a set of new cost functions to enhance theadaptability of the algorithms, according to HILS and Road Traffic Testbed atIntelligent Transportation Systems (ITS) Research Center, Wuhan University ofTechnology.First, the Similarity Theory is introduced, which covers the geometricalsimilarity, kinematic similarity, and dynamic similarity between theoretical model andscale vehicle. The theory provides a parameter calibration method to modeling scalevehicle with intelligent experimental platform. Regarding the road geometrymodeling, a function of clothoid is used to develop both curve and slope models; inthe vehicle kinematics modeling, the Lagrangian method is used to develop alongitudinal and lateral coupling model with3DOF, and it considers longitudinal,lateral and yaw motion. Meanwhile, the road traffic testbed and scaled vehicle areconnected to build the experimental platform of this study.Secondly, the proposed adaptive vehicle speed control is examined for bothslope and curve segment. Under the slope scenario, the slope angle and its rate arederived. According to the rule of energy loss of a dissipative system, the adaptivevehicle speed control can be converted to H∞control with an optimized storagefunction. When the performance criteria of γ dissipation is set up in the speedcontrol system, the HJI (Hamilton-Jacobi-Issacs) method is used to develop a storagefunction including the excitation cost of the slope geometry and the compensationcost of speed and the longitudinal acceleration change. A Backstepping method is used to successively approximate the speed control system with the γ dissipationinequality along the forward transmission of the integrator chains. Simulations forspeed adaption on the up and down slopes are carried out, and it is found that theproposed adaptive speed control rules can not only regulate the speed fitting with thechange of the slope angle, but also explicitly consider driving safety and energy loss,which is a significant merit when compared to those without considering such effects.Third, under the curve scenario, a longitudinal and lateral coupling model with3DOF is used to simulate the lateral deviation from the road center line under the curvespeed, in order to study an optimal longitudinal and lateral control. Once thecurvature and its rate is model, the adaption control of lateral deviation is calculatedwith the sum of least square between the observed and theoretical lateral deviations,on the basis of the γ dissipation of the speed control system. The LMI method is usedto compute the positive definite matrix, which ensures the lateral deviation adaptionsystem maintains the Hurwitz stability. The simulations for both adaptive speed andlateral deviation on the curve are accomplished and it is found that the speed adaptionlaw developed for the curves can not only regulate the speed and the lateral deviationfitting with the change of the curvature, but also explicitly consider driving safety andenergy loss.Finally, hardware-in-the-loop based tests for the adaptive vehicles are carried out,which covers both speed adaption on slope and lateral deviation adaption on curve.The feasibility and effectiveness of the proposed control methods are found to beacceptable over the tests.