| GPS Q&A
By Bill Elliot Q.My company is developing an enterprise-wide GIS. One planned application is vehicle tracking and fleet management. What factors need to be considered when developing a GIS that will support GPS-based vehicle tracking applications?
W.S. Cleveland, Ohio A.Just as GIS databases come in all shapes and sizes to support differing user needs, the development of digital background maps to support GPS-based vehicle tracking and fleet management applications can vary based on specific user needs. First, some questions need to be answered about the nature of planned vehicle tracking applications before GIS requirements can be defined:
1. How accurately do vehicles need to be located/tracked? To the nearest house address, street block, section of town, or major city?
2. Is the purpose of the vehicle tracking application exclusively to provide visual location of vehicles on a dispatch center digital map screen? Or, are more sophisticated applications planned?
3. Will building/house addresses and/or street address ranges be used for directing vehicles to specific locations?
4. Are routing and/or scheduling applications planned as part of the fleet management system?
5. Will vehicles be dispatched to company-managed outside facilities, such as sending a repair vehicle to the location of a broken water main?
6. Will in-vehicle, real-time map display screens be supported as part of the application?
7. How often will vehicle locations be updated on the dispatch center's map display? Will vehicle location updates be frequent and automatic, or will dispatch center 'polling' be used to initiate vehicle location updates on the dispatch center's map display? Differential Versus Non-differential
As a starting point, the scale at which vehicles need to be tracked to support your applications will determine whether or not differential GPS techniques should be used. If the application locates vehicles traveling across a multi-state area, for example, and you need to only know in what town the vehicle is located (question #1), non-differentially corrected GPS might suffice to track and locate vehicles. Long haul trucking companies, for example, may not require differential GPS to track their trucks.
Many applications, however, call for 'street-level' mapping of vehicle locations. In these scenarios, differential GPS is almost always required to provide the accuracy needed to view vehicle tracks on or near street centerlines of the underlying background maps. Specialized commercial data suppliers who provide digital street centerline data specifically for vehicle tracking applications typically specify their data at +/- 15 meters absolute accuracy. Differential GPS techniques are required to consistently maintain vehicle tracks within the accuracy range of such underlying street centerline maps. Real-time or Near Real-time
If differential GPS is to be used for the application, it can take place in either real-time at the vehicle's GPS receiver or in 'near real-time' via post-processing at the dispatch center. Your answer to question #7 will help determine whether or not near real-time post-processing is acceptable. If vehicle locations are updated every five minutes or on an as-needed basis through polling, then near real-time differential post-processing is an option. If in-vehicle map display technology is to be used (yes to question #6), then real-time differential is the preferred method, since it will provide real-time location updates in the vehicle while driving. As real-time differential GPS equipment continues to drop in price, this option will become the preferred solution for most vehicle tracking applications in the future. Differential GPS on a Raster Base
Let's presume your vehicle tracking application is solely for purposes of providing a visual map display of vehicle locations at the dispatch center with no other applications planned (yes to question #2). In this case, a raster GIS background map may be sufficient to support differentially corrected GPS-based vehicle tracking.
The raster GIS approach is highly dependent upon the display scale of the background map to be used. It has been my experience that, when properly geopositioned using GPS control points, USGS 7.5 minute quadrangle maps or commercial map atlases at 1:24,000 (1 inch equals 2,000 feet) offer an adequate backdrop for tracking vehicles along street centerlines to the nearest street block using differential GPS. If a raster GIS background is to be used in this manner, it is essential that a sufficient number of control points be GPS-mapped on the ground, referenced to the raster map image and used to warp the background map into its approximate position in the GPS reference grid. When positioning multiple scanned map images in this manner, overlaps and/or gaps may occur at map edges. In some cases, commercial map atlases must be warped to such an extent as to render the map's annotation difficult to read. The resulting map image, however, is usually of sufficient accuracy to support GPS-based vehicle tracking applications.
In one example, a USGS 7.5 minute quadrangle sheet was scanned, georeferenced through a warping process, and used as the background map for GPS-based vehicle tracking. When viewed at a scale that showed one quarter or more of the quad sheet on the display screen, the vehicle's GPS track visually appeared to overlay the map's street centerlines precisely. However, upon 'zooming in', the same vehicle track appeared noticeably offset to one side of the same street centerline. In this case, the dispatcher rarely needed to zoom in further than to view a quarter of the quad map, so the vehicle's track-to-map resolution was sufficient for the application. Street Level Accuracy
Many vehicle tracking applications require street level accuracy. For example, if vehicles will be dispatched to specific building addresses or city street blocks, as is the case for most fire, police or customer service providers, then vehicles need to be tracked at street address level accuracy (yes to question #3). For these applications, an intelligent, vector-based GIS background map will be required.
Many vehicle tracking applications incorporate geocoded address databases that support the rapid location and display of a specific house address on a map simply by entering the address into an 'address finder' program. Since geocoded address databases are very difficult and expensive to develop and maintain, most user organizations opt for the purchase of off-the-shelf geocoded address databases from commercial suppliers. The majority of geocoded address data sold commercially today was derived from TIGER file data created and collected by the U.S. Census Bureau. TIGER files are rarely used in their native form to support GPS-based vehicle tracking applications, but are the basis of many enhanced data products offered by commercial data suppliers. These improved and enhanced versions of TIGER are best suited for vehicle tracking applications.
Frequently, 'map matching' techniques are used to improve the effectiveness of TIGER-derived data for GPS-based vehicle tracking applications. In map matching, software automatically moves and repositions a vehicle's GPS track to the nearest road segment for display purposes. Be advised that TIGER-based and other 'street level' data sets from commercial data suppliers tend to be more positionally accurate and complete for the large metropolitan areas. Data tends to be incomplete, inaccurate or non-existent for many rural areas. When considering the purchase of commercial data, be aware that the data supplier's release of new construction subdivision and street data may take several months. Routing and Scheduling
If your application will involve routing and scheduling (yes to question #3), then a more robust street and address database will be required. Use of non-attributed street centerline data is usually insufficient for supporting effective routing and scheduling applications. For these applications, not only must the street centerlines be accurate (+/- 15 meters), but additional road segment attributes such as turn restrictions, one-way streets, and average traffic times/speeds (impedance) will need to be included in the database. A small group of commercial suppliers provide GIS databases specifically tailored for routing and scheduling applications. These data sets can be customized and enhanced to meet unique routing and scheduling requirements. Design, Engineering and Construction
If your application requires that vehicles be dispatched to outside facilities, such as telephone, water, sewer, gas or electric infrastructure (yes to question #5), then the GIS should support the GPS positioning of facility features on a seamless base map. Merging vehicle tracking (AVL) and outside facilities management (AM/FM) technologies can provide significant benefits for trouble call applications that require quick identification and location of damaged facilities, followed by rapid dispatching of the nearest available field crew or service vehicle to the trouble call site. Emerging AVL/AM/FM applications will significantly improve trouble call response times and overall management of utility field forces in the future.
Difficulties encountered implementing AVL/AM/FM solutions more often than not are caused by inadequate AM/FM map databases rather than the lack of an innovative AVL solution. To support facilities-based AVL, the underlying map base (AM/FM/GIS) must provide a high degree of "absolute" geographic accuracy that will relate mapped facility locations to the GPS location of vehicles. In general, the legacy of AM/FM systems tends to be "drawing-based" in that the seamless map is created from an amalgamation of drawing sheets (work prints, plats, engineering details, schematics, etc.). Each drawing sheet may have high "relative" accuracy and include very precise dimensioning. However, many of the drawing sheets from which AM/FM maps are created lack geodetic control reference points or grid projections. When seamed together on a geographic base, drawing sheets may not edgematch well. Map distances and directions may be noticeably skewed across large, seamed-up areas. Individual drawings need to be positioned across a seamless base map using GPS control points in order to support GPS-based vehicle tracking applications. Many AM/FM databases are not being developed in this fashion. Combining GPS Vehicle Tracking and GIS Data Collection |
The capabilities of GIS and GPS-based vehicle tracking are rapidly being combined to offer new and exciting applications for improving the efficiency of the mobile work force. New GPS/GIS systems, such as PowerTrak (from GeoResearch), provide sophisticated GPS-based fleet management and field mapping capabilities in a single, open systems environment. These 'combined' systems make it easier to build and maintain GIS databases that can support robust GPS-based vehicle tracking applications. The creation of GIS databases that meet the needs of a wide range of emerging and increasingly popular GPS-based vehicle tracking applications will be the real challenge facing our industry as we move into the 21st century. Back |
