Research On Trajectory Planning And Control Algorithm For Parallel Parkiag

Traffic congestion, vehicle scratches and other issues caused by curb parking lack ofstandardization has become a major problem for traffic management. To this end,automatic parking has become the research and development emphases of the vehicleactive safety technology. Curb parallel parking can be divided into forward-parking andbackward-parking according to parking space.The trajectory planning based on B-spline and multi-arcs are proposed forforward-type parking considering the constraint of space. For backward-type parking, a carparking control concept for real-time application paper presents is presented under thephysical limitations and non-holonomic constraints. And a feedback controller is designedfor the trajectory error system.The paper is research result of sub-topic of road safety reasoning research based ontraffic situation assessment, which is province natural fund. The main and innovatoryresults are as follow, for forward-type parking:(1) According to Ackermann steering geometry, the kinematics model is createdfrom the simplified model of a vehicle under the conditions of relatively abundant parkingspace. The non-linear multi-constraints are built by analyzing parking environment andvehicle constraint. The trajectory curve based on B-spline is obtained by solving themulti-constraint equations using curvature minimized as single aim to plan trajectory byone step for forward-type parallel parking.(2) The parking method based on multi-steps forward and backward parking isproposed for narrow space. The trajectory curve is composed of multi-tangent arc. Therelevant constraint equations are constructed by analyzing potential collision existing in theenvironment, and the trajectory parameters are obtain by using parking space minimizedand curvature minimized as single aim respectively.For backward-type parking:(3) A trajectory generator consisting of a novel fast geometric path planner and anoptimization based solution for the path-following problem is presented. The geometricplanner consists of a series of static optimization problems. Obstacles are described by inequality constraints based on the Minkowski sum. The path-following problem exploitsthe flatness property of the car enabling the reformulation of the system dynamics in formof a linear second order differential equations.(4) Additionally, a trajectory tracking feedback controller is presented. To this end, theerror system is transformed into a chained-form system. The control inputs are calculatedusing an integrator back-stepping method and the stability of the closed-loop system isproven utilizing Barbalat’s.