Executive Summary

Executive Summary

Connected Vehicle applications require established methodsof roadway feature representation and reference in the form of a map. Numerousmapping methods exist to acquire and record geospatial data that representroadway relevant objects and terrain. Based on our extensive analysis carriedout in this research project, the preferred existing acquisition method forroadway mapping will likely utilize GPS/IMU-integrated LIDAR sensing thatgenerates a three dimensional georectified point cloud data set. When dataacquisition is conducted with multiple integrated sensors at a high frequency,the resulting dataset will be sufficiently large to require automatedapplication specific data processing for roadway relevant feature mapping. Thisis particularly true when mapping at the national or global scales necessaryfor commercial success.  

This report entitled “Best Practicesfor Surveying and Mapping Roadways and Intersections for Connected VehicleApplications” presents a technology and methodology review of current mappingmethods and technologies. The main body of the report is divided into six mainchapters: 

[if !supportLists]1.     [endif]MappingMethodology Assessment;

[if !supportLists]2.     [endif]MobileMapping System Enhancement;

[if !supportLists]3.     [endif]MapRepresentations;

[if !supportLists]4.     [endif]MapRepresentation Updating;

[if !supportLists]5.     [endif]FeatureExtraction Methods, and;

[if !supportLists]6.     [endif]IntersectionMapping Experimental Results.

Subsequent chapters discuss best practices, conclusions, andfuture work. The report also includes a list of references and threeappendices. 

Connectedvehicles require accurate and up-to-date maps both to allow coordinationbetween vehicles and with the infrastructure. Such maps may also have utilityfor application aspects such as vehicle position estimation or control. InChapter 1, Mapping Methodology Assessment, we describe several ways that mapscan be acquired. Based on the analysis, it was found that mobile terrestriallaser scanning (MTLS) methods work best for connected vehicles purposes. Theresearch team has previously participated in the development and operation of aMobile Positioning and Mapping System (MPMS) deployed and tested at TurnerFairbanks Highway Research Center. This system meets a number of key criteriaincluding accuracy, robustness, efficiency, cost, safety, and usability.Chapter 1 reviews the MTLS approach and examines the mapping and positioningaccuracy requirements of a large number of CV applications, particularly thoseapplications listed in the Connected Vehicle Reference ImplementationArchitecture (CVRIA).

MPMS’s are mounted on a vehicle platform which collectspositioning and mapping data from a variety of sensors and combines them toprovide accurate, and continuously available information about both thetrajectory of the MPMS and the surrounding areas, yielding more accurate andprecise location detail and associated feature maps. This is achieved through acombination of global positioning satellite (GPS) technology, feature-basedaiding sensors (vision, RADAR, LIDAR) and high-rate kinematic sensors (wheelencoders or inertial measurement units (IMU)) to capture and process multiplelocation and feature-based signals and to bridge data gaps whenever sensorreception is interrupted. The improvement to the UCR MPMS hardware and softwareis the focus of Chapter 2.

For successful collaboration with automakers, it is expectedthat some entities (government or commercial) will develop and maintaincontinent-scale roadway map databases, and eventually global scale. Maintenanceof this master map will result in differences between the master map and themaps stored on user vehicles. The master map is too large to be convenient forwireless communication to users in its entirety; therefore, mechanisms havebeen defined for communication of application relevant pieces of the map toconnected vehicles. Chapter 3 discusses the processes, general standards, andthe SAE J2735 standard, which along with its modifications for demonstrationpurposes is the dominant standard for connected vehicle applications.

It is certain that the infrastructure and roadway featureswill change over time, particularly for corridors that are heavily utilizingconnected vehicle technology. Therefore, once a map database is established (asdescribed in Chapters 2 and 3), a key issues is how it can be updated toaccommodate changes in the infrastructure, the introduction of new mappingtechniques, or the desire to map additional features. Chapter 4 brieflydescribes different possible update technologies and approaches.

Chapter 5 presents an automated feature extraction approachexplaining the data processing steps utilized to transform a georectified pointcloud representation of the roadway environment to relevant intersectionfeatures represented in a SAE J2725 map message. SAE J2735 is the dominantcommunication media and associated map message format intended to representintersection geometry and features appropriate for connected vehicleapplications. The feature extraction methodology presented is intended toexemplify a uniform approach applicable to standardized intersections meetingaccepted roadway design criteria. As such, the feature extraction methodologycan serve as a template for feature extraction beyond the scope of J2735 basedapplications. 

The roadway feature extractionprocess consists of the following primary steps:

    Preprocessing to extract the georectified pointcloud and associated MPMS trajectory portions relevant to an intersection thatis of interest;

    Identification and extraction of the roadsurface point cloud, road edge curves, and median edge curves;

    Conversion of the intersection road surfacepoint cloud to an image to enable feature extraction using image processingtechniques;

    Image-based roadway feature extraction; and

    Translation to a J2735 intersections feature mapoutput format.

The feature extraction and map representation methodologyare described with a detailed explanation of data processing and integration,including examples. This detailed approach allows for future feature extractionof relevant roadway features in a connected vehicle environment. Theperformance of the semi-automated feature extraction approach is demonstratedusing eleven example intersections along El Camino Real in Palo AltoCalifornia. The report concludes with recommended practices, suggestions forfuture work, and conclusions.