Space control law design in adaptive cruise control vehicles using model predictive control

摘要

Driver assistance systems (DAS) are being broadly utilized in view of the fact that driver workloads will be decreased in DAS-equipped vehicles. A variety of DASs have been designed, and adaptive cruise control (ACQ, as a type of DAS, is the subject of the present paper. The ACC designers' main concerns are to provide comfort, which is characterized by minimized jerky movement and driver workload, and safety, which is characterized by the ACC system's collision avoidance (CA) capability within its region of operation (determined on the basis of rigorous discussions derived from the relevant literature). Furthermore, ACC designers typically use more than one control law for ACC-equipped vehicles because the ACC must react differently to variations in driving conditions. In view of these points, it may be seen that, in utilizing ACC, the driver's operating workload will be more that of a supervisor than an operator, owing to the longitudinal automation ability of ACC. In addition, an on-line control law can operate a vehicle in all driving conditions. The space control law developed in the current paper explains the compromise between safety and comfort. ACC controllers are, typically, designed on two levels to increase the system capabilities: the upper level and the lower level. The upper-level controller comprises speed and space control laws, the latter of which computes the reference acceleration profile on the basis of the measured range, range-rate, and speed and acceleration of an ACC-equipped vehicle. Following reference acceleration, the desired range and range-rate will be achieved. The lower-level controller controls throttle and brake actuators to track the reference acceleration profile. In this paper, the desired acceleration in any time step is obtained by formulation of a constrained optimal control problem (QP) and employing a receding horizon control (RHC) strategy. Three baseline scenarios and the high-acceleration/deceleration driving cycle LA92 are considered to evaluate the control law. Note that the model predictive control (MPC) algorithm shows an advantage over other controllers in terms of safety and comfort. All working steps are executed using Matlab software.