Research On Micro Intelligent Vehicles Based Intelligent Transportation System Simulation Platform

The urban transportation system is a typical large scale complex system. Trafficcongestion and the relevant problems have increasingly become the worldwidepressing issues. Intelligent transportation system based scientific management andcontrol of traffic flow is considered as one of the most cost-effective strategies toresolve the above problems. One of the initial issues of the realization of intelligenttransportation system is the modeling and simulation of traffic flow. Some of theresearches are carried out based on software simulation. While, software simulation isinsufficient to reflect the reality and hard to simulate human driving behavior. And theexperiments in actual traffic system are usually complex and time-consuming. Alsothere hardly ever exists chances for multi-intelligent vehicles to be involved in. As aconsequence, a traffic flow simulation approach based on multi-micro intelligentvehicles is proposed which can be freedom from law, weather and other factors whichtroubles experiments in actual traffic environment by creating a micro proportionaltraffic flow environment. Meanwhile, some other problems will be produced andrequired to be resolved.Firstly, because of lack of global position feedback like GPS equipment, thelocation and navigation approach in micro traffic environment should be researched.In this paper, a location and navigation approach based on hierarchical topology mapsis proposed to realize the lane-class location in micro traffic environment. And due tothe reason that some driving routes (e.g. turn-left and turn-right) might be forbidden incertain intersection though the lanes are connective, which is hard for traditional pathplanning approach to achieve a faultless shortest path. In this paper, an improvedglobal path planning approach is proposed based on the classical Dijkstra pathplanning approach to give a result with the consideration of traffic rules in everyintersection.Secondly, due to the limitation of the hardware configuration of micro intelligentvehicle, the environmental perception approaches should be studied to save thecomputational resource and optimize algorithm in order to enable the micro intelligentvehicle be qualified with real-time characteristics. An improved visual selectiveattention method is applied in traffic lights and traffic signs detection for microintelligent vehicle based on the residual spectrum theory which can decrease theregion of interest for environmental perception. Also an improvedSpatial-Bag-of-Features approach is applied for traffic sign recognition to meet thereal-time requirement. Experiments show that the proposed method is suitable for the environmental perception of micro intelligent vehicle.Thirdly, to simplify the control system design of micro intelligent vehicle, acontrol system architecture for two-wheeled mobile robots based on multi-objectiveflocking is applied in micro intelligent vehicle in order to realize the different controlstrategies in various autonomous driving tasks. Experiments show that the controlsystem architecture of micro intelligent vehicle can support the basic autonomousdriving tasks in actual traffic environment.Finally, in the field of autonomous parking research, to overcome the problemthat the traditional parking area detection based on distortion correction andorthographic projection cannot search the parking area in a remote distance andrequire the ground to be smooth, a parking area detection method based on contourdetection and matching is proposed. Then, histogram based parking corner detectionis carried out to achieve the relative position between the micro intelligent vehicle andparking area. Next, a local path planning method is carried out to obtain the parkingtrajectory. Experiments show that this method is suitable for the autonomous parkingtask of micro intelligent vehicle.As a consequence, this micro intelligent vehicle is capable of the simulation ofdifferent autonomous driving tasks in reality and is suitable for the research of trafficflow simulation and intelligent transportation system.