Path design and receding horizon control for collision avoidance system of cars

摘要

The paper deals with path design and control realization problems of collision avoidance systems (CAS) of cars (ground vehicles). CAS emergency path design is based on the principle of elastic band with improved reaction forces for road borders and static obstacles allowing quick computation of the force equilibrium. The CAS path (reference signal) is smoothed and realized using receding horizon control (RHC). The car can be modelled by full (non-affine) or simplified (input affine) nonlinear models. The nonlinear predictive control problem is solved by using time varying linearization along appropriately chosen nominal control and state sequences, and analytical solution of the minimization of a quadratic criterion satisfying end-constraint. Differential geometric approach (DGA), known from control literature for the input affine nonlinear model, has been used for control initialization in the first horizon. For state estimation Kalman filters and measurements of two antenna GPS and Inertial Navigation System (INS) are used. A stand-alone software has been been developed using the C Compiler of MATLAB R2006a satisfying real time expectations.