Research On Control Of Lane Depareture Assistant System Based On Machine Vision

In order to reduce the traffic accidents caused by human errors,the intelligent vehicle driving assistance system has become a hot topic in academic and business fields in recent years.Lane departure assistance system is a relatively popular driving assistance system on the automobile at present,which can remind the driver that the vehicle is about to deviate from the lane,assist the driver to achieve lane tracking and actively prevent lane departure,improve the driving safety of the vehicle,so it is increasingly paid attention to.This dissertation first reviews the research background and development history of the vehicle lane departure assistance system.In this process,it focuses on the development background and practical application of lane departure assistance systems.It focuses on the classification of the application of different control methods for lane departure assistance systems.The functions and principles of various lane departure assistance systems are summarized.The research status of lane recognition algorithm,lane departure assist control,man-machine coordination decision-making method and vehicle dynamics control are summarized.In this dissertation,the theoretically and experimentally researches focused on the key technologies of the lane departure assistance system,including the lane departure detection,lane departure assist control and man-machine coordination control and so on,have been done from the following five aspects.(1)On the basis of the recognition of the straight lane line,using the method of projection statistics,a curved lane line recognition method based on hyperbolic model is proposed.In addition,In addition,in order to obtain the curvature of the curved road more accurately,a curved lane recognition algorithm based on near-field straight lane line information for the curvature of the edge of the region of interest is proposed,this algorithm first in the near field of lane straight line identification,the direction and location informationof the lane straightlineas the initial parameters of curve fitting.In the region of interest in planningcurvature change area,the curvature of the curve extending to the line edge point is calculated bycombining the ROI edge points in pixel coordinates with the current near field straight linear equation parameters,then the curvature interval that the curvature calculation value belongs to is voted.Gets the current lane curvature based on the most voted for curvature interval.Finally,the equation ofcurve lane-line is obtained by combining the curvature value with the line parameter information.(2)Firstly,considering the influence of vehicle state on Lane departure assist control,two variable parameters of vehicle longitudinal speed and lateral vehicle speed are introduced to construct road potential field function.The trajectory prediction theory is used to determine the design parameters of potential field function,and the potential field gradient is used to solve the desired steering angle.Secondly,the state equation model of path tracking error variable with variable parameters is established,and the stability of the closed-loop control system is analyzed.Then,on the basis of considering the influence of vehicle state on lane-keeping control,taking driver’s torque and intentional path as input,the weight of auxiliary control is dynamically adjusted by using fuzzy rules to balance the problem of man-machine coordination inLane departure assist controlprocess.Finally,an auxiliary Lane departure assist control method considering the influence of vehicle state change and man-machine coordination is realized.(3)In the research ofIn the study of lane departure assist control,not only consider the road tracking and ride comfort,but also think about the steady of vehicle about tracking target path,parameter variationand as well asexternal interference when both the controller and driver control the vehicle.In this paper,consist of steering system modeland 2-freedom model,based on both vehicle state and expected angular rate.The disturbance inputs are cross wind and pilot control.The feedback variables are yaw rate,hand wheel angle,measured values of torque sensor and desired yaw rate.By considering the effect of vehicle parametersand sensornoise,the layerμintegrated controller is designed to achieve the desired lateral displacement of vehicle tracking as well as make sure that the vehicle can track the target path stably.(4)At present,the lane departure control is implemented by independent control of steering control and differential brake control.A new control strategy is designed to coordinate the two control modes.The warning threshold is designed according to the response of vehicle steering system and a multistage warning system for the vehicle deviating is designed based on the time-model of cross-channel.Furthermore,in order to guarantee vehicle stability,using model predictive control algorithm,a multistagewarning systemis added the reasonableconstraint.By designing differential braking and activesteeringcontroller,the strategy of preventinglane departure based on active steering and coordinated control of differential braking is realized.In the process of coordinating control,Vehicles can drive more safely and smoothly as well as consider to the human-machinecoordination.(5)Using the hardware-in-the-looptest bench designed by our research group,verify the steady control of vehicle lateral motion,the human-machinecoordinated control of the lane departure control and coordinated control based on active steering and differential braking.On embedded platform develop lane line detection system,and using laboratory’s active steering controller and active baking controller that are used to realized vehiclebottom control.On this basis,the effectiveness of the proposed the lane departure control method is verified according to build vehicle test platform.