Reverse Parking Control Method Based On Neighborhood System And Soft Optimization

At present, intelligent vehicle become a hotspot in the research of intelligent products which have attracted widespread attention of automobile manufacturers and IT companies. Considering the difficulty to achieve full intelligence, coupled with the privacy and the wide gap between theoretical research and application, to make a research on the partly intelligence, such as self-parking which is a key technology to achieve, has an important significance in the short term. It is also one of the research focuses of this paper.Firstly, the relation between fuzzy soft sets and generalized multi-objective optimization is theoretically studied. The thought is to take the membership functions of fuzzy soft set as the objective functions of multi-objective optimization problem. The membership functions are established from different single attributes to give things approximate descriptions based on fuzzy soft sets. It obviously makes complex issues simplified and unfortunately the solving method is still not satisfied, while the multi-objective optimization problem has systematical solving method. So we try to give the transformation between the two methods. And a method to transform a multi-objective optimization with infinite number of solutions into a finite partition optimization from a practical point of view is presented, which is theoretically proved that it will approximate the optimal solution of the optimization problem at any accuracy only if the parameter domain is divided thin enough.Secondly, the thought of soft optimization is applied to select satisfactory feasible neighborhood and satisfactory control rule in the feasible neighborhood based on neighborhood system. The dynamic decision process can effectively mimic behavior of intelligent lives, and the main points of the decision process are as follows:selecting a suitable neighborhood system according to the dynamic characteristics of the controlled system and implementing periodic control. The process consists of two steps. The first step is to find the satisfied feasible neighborhood in the neighborhood system by the method of soft optimizing according to the current state and regard it as static; second is to give the control variable corresponds to the trajectory in the satisfied feasible neighborhood and implement control action by using multi-objective soft optimization. The whole control is composed of the sequence of each neighborhood control process.Then the intelligent vehicle’s dynamic characteristics are analyzed. And the dynamic reversing process is proved to be a typical monotonous inertial system. Based on the initial position information, seven standard reversing positions are selected. By this, seven corresponding standard rules are determined to constitute a group of control rules. In connection with the monotonous and inertial natures, comprehensive control functions can be implemented to realize reverse parking by any initial positions.Finally, simulations on several different representative initial positions in the actual standard garage by Matlab verify the feasibility of the method. Compared with other algorithm, the method in this paper is simpler in calculation and can decrease the configuration requirements of a smart car.