Research On Longitudinal Speed Control For Autonomous Vehicles Adaptive With Road Grade And Vehicle Mass

The disturbances from driving environments will affect the vehicle longitudinal speed control during the vehicle longitudinal driving, such as the change of road slope, the change of vehicle mass, the change of wind resistance, the change of road resistance and so on. Considering the above disturbances, it is important to make the controller more adaptive with the disturbances. For solving the above problem, researchers have proposed many controllers adaptive with the disturbances. This paper introduces a proportional inner model controller adaptive with disturbances, which has been applied in the real autonomous vehicles. This controller has an inner model structure, which can get the disturbances of the environment. By compensating the disturbances, the controller can be adaptive with the disturbances. But the disturbances got form the inner model structure cause by many factors, which have different characters. For example, the road grade changes fast with time, but the vehicle mass mainly keep the same during driving. So it is difficult to select one coefficient for the filter to filter the disturbances. For making the controller more adaptive with the disturbances, we can design estimator to estimate the different disturbances. But the different characters of the disturbances make it difficult to design estimators. In addition, vehicle is a highly no-linear system with complex subsystems. It is also crucial to build a vehicle model, which can be applied in reality.Based on the above analysis, the objective of this paper is to design longitudinal speed controller for autonomous vehicles adaptive with road grade and vehicle mass. For this purpose, this paper firstly built a longitudinal vehicle MAP, which is suitable for application in reality. This longitudinal vehicle MAP can not only describe the vehicle longitudinal dynamics during the acceleration but also describe the vehicle longitudinal dynamics during the braking, which is different from the traditional vehicle model, for that is more suitable in reality. Then this paper established a simultaneous estimator of road grade and vehicle mass. Comparing with the exiting estimator, the simultaneous estimator designed by this paper can estimate the road grade and vehicle mass not only during the acceleration but also during the braking. Based on the above work, this paper proposed to build a longitudinal speed controller for autonomous vehicles adaptive with road grade and vehicle mass. Comparing with the proportional model controller, the controller designed by this paper can estimate the road grade and vehicle mass, which are the feedback of the controller, for that the controller can be more adaptive with the disturbances.This paper gives efficient experiments for the above work. Firstly, we test the longitu-dinal vehicle MAP by vehicle dynamics software veDYNA and real autonomous vehicles. Then we test the longitudinal speed controller adaptive with road grade and vehicle mass designed by this paper by comparing with the original proportional inner model controller.In the last, the future research of this paper mainly include to improve the longi-tudinal vehicle MAP during the braking and to estimate more disturbances such as the road coefficient to make the controller more adaptive with the disturbances.