Exploring Local
Mike Dobson of TeleMapics on Local Search and All Things Geospatial

3D Road Geometry Requirements for Automotive Safety and Energy Management Applications

November 22nd, 2009 by MDob

Well, I’ve finally returned from my working vacation in Europe. Seventeen days and I only saw the sun for about twenty minutes, which is hard to take for a resident of Southern California. On the other hand, Austria and Germany are beautiful at this time of year and we had most of the attractions we visited to ourselves, so the timing was a good choice.

As I mentioned in my last blog, I was in Europe to chair a session hosted by Intermap Technologies at the TeleMatics Update conference in Munich. As you may know, Intermap is a company that produces digital elevation models, orthorectified radar images and a host of geospatial enabled products based on their proprietary, airborne Interferometric Synthetic Aperature Radar (IFSAR).

Over the past year, Intermap’s Automotive Division has been focusing its efforts on the application of their technology to solving problems of interest to the automotive industry. The company has been making nice progress in licensing its accurate 3D terrain and elevation data for 3D In-Dash Visualization purposes. However, it is my guess that the biggest opportunity for the company is in the areas of energy management and vehicle safety (better known in the industry as Advanced Driver Assistance Systems).

As you have probably guessed by now, Intermap has come up with a proprietary process to extract 3D Road Geometry for road networks from its IFSAR-based imagery (i.e. extracting from their high-resolution, controlled imagery database of coverage the horizontal position of a point on a road and the elevation of a road centerline at that position in a highly accurate manner). This 3D Road Geometry database is the asset that Intermap hopes will help it become a leading supplier to those companies in the automotive industry interested in advanced energy management and safety applications.

It is important to note here that while the world seems to be focused on Google’s recent announcement of its proprietary navigable map database for the United States, the ADAS and Energy Management markets are focused on map data that are created at significantly higher levels of positional accuracy (both horizontal and vertical) than commonly found in data supporting routing and navigation applications. For example, Navigation and routing assume that the vehicle must be located on and operating on a road and depend on map-matching algorithms to rectify the situation when it appears that cars are not being driven on a road. In a simplified sense, if the GPS reading describing the position of a car is not on a road described in the map database, map-matching will find the road or street described in the database that is nearest to the car, assume the car is positioned on the road (until a GPS reading or maneuver leads it to change that assumption) and proceeds with its navigation or routing calculation. Although NAVTEQ and TeleAtlas are very tight lipped about the accuracy of the navigation databases, the general agreement in the industry is that the positional accuracy is within plus minus 15 meters (absolute) and plus or minus 5 meters (relative).

Advanced Energy Management and Automobile Safety Applications require a significantly higher degree of accuracy in the map databases used to support their mission critical functionality. Although the industry has not agreed on a standardized definition, it appears that many applications are focusing on horizontal accuracies of plus or minus 3 meters or less (absolute) and vertical accuracies better than 1% of the distance traveled (rise/run). As a consequence, the mapping industry is beginning to tool-up new methodologies for capturing road geometry and elevation data at higher levels of accuracy than in the past. (For more information you might be interested in reading my article on ADAS and 3D Map Databases that can be found here.)

Navteq, for instance, is reputed to be driving roads using vans equipped with inertial navigation units and other precision measurement devices to collect the data needed to meet the needs of ADAS and Energy Management applications. TeleAtlas is relying on crowd-sourced probe data to collect high accuracy, positional and elevation data, while Intermap has opted to use precision remote sensing as the source of its high accuracy 3D Database.

What is intriguing to me is that gathering this new generation of data will involve clearing at least two backbreaking hurdles. First, is the issue of coverage. For instance, it took nearly two decades for NAVTEQ and TeleAtlas to provide a relatively comprehensive, reasonably complete road databases in order to support the target markets for their navigation databases (roughly speaking this means covering Europe (the EU), the United States and sections of Canada). The second hurdle, as noted is the ability to create map data across these large coverage areas that uniformly meets the accuracy needs of the industries comprising the target markets for these data.

While NAVTEQ’s approach to gathering 3D map data is tried and true, it is also slow and expensive. Using this methodology, NAVTEQ will likely focus its coverage efforts on the road classes that move the most traffic, but at the sake of comprehensive coverage of all roads. For example, it has been rumored in the industry that NAVTEQ does not plan to collect all road categories (e.g. those that NAVTEQ classifies as Functional Class 5 – see this official website for more details on the NAVTEQ classification). Road functional classes are a construct developed by NAVTEQ, although the term is now widely used in the navigation industry, to describe both the mix of road types in a transportation network and their numerousness. While the number of functional road classes used to describe a transportation network varies between map companies, NAVTEQ employs five functional classes to describe all roads.

ALthough I have no specific numbers I can quote you, it is my belief that the relative frequency of road classes in Western Europe and the U.S. chart as shown below. Not collecting accurate 3D data for roads in, say, Funtion Classes 4 and 5 makes it impossible to provide safety and fuel efficiency applications that could be applied across the geographies of these markets. NAVTEQ has yet to decide how to approach this challenge, but it is likely that driving all the roads in its feature classes would take many years and retard its success in the market.

Relative frequency of road class categories in the Europe and the United States.

TeleAtlas appears to be depending on crowd sourced data at the expense of field collection. While crowd-sourced data can provide many advantages in collecting attribute data for road vectors in a navigation map database, it is less clear that crowd sourced data will ever meet the positional accuracy and elevation requirements needed to define road vectors for safety and fuel efficiency applications. A second problem with crowd sourced data is that it appears to be spatially autocorrelated with population density. Where you have large agglomerations of people, you will likely receive adequate input from the crowd. Conversely, it is less obvious that crowd sourced data will provide any significant contributions in rural areas since crowd sourced observations may be too low in volume to provide reliable data.

Intermap has taken an alternative approach by using remotely sensed data and appears to believe that its production platform can provide both the accuracy and coverage (including all road classes) that the markets require and to be able to accomplish this feat over a relatively short time frame. The company has been quietly developing its 3D Map Database product at its facility in Jakarta, Indonesia and Intermap has recently released its database of Germany for testing by interested parties.

Hopefully, the information above helps you to understand why I was interested in accepting Intermap’s invitation to chair a session at TeleMapics Update on “3D Road Geometry Requirements and Advanced Automotive Applications.” The development of databases of the quality and coverage to support both automotive safety and energy management applications is the next big test for the mapping industry and I, for one, am very interested in seeing how it will play out.

The panel assembled to discuss the topic included myself as moderator, Eric DesRoche, Senior VP of Automotive (Intermap), Christian Ress, Technical Expert, Telematics & Navigation (Ford), Christopher Wilson, Director of Strategic Research (TeleAtlas), Wieger Engbrenghof, Project Manager TeleMatics (Continental). Navteq was invited to participate, but they declined. In essence, the panel consisted of two representatives from companies interested in supplying 3D Map data, one automotive OEM, and one Tier 1 supplier to the automotive industry, which, I thought, would provide a good mix of opinions across the 40 minutes available to us for discussion. In addition, Christian Ress from Ford is the new chairman of the ADASIS Forum, which added to the value of the deliberations by the panel.

I presented the panel with a series of 5 questions of interest to me (and, I think, to the industry). Next time I will present the questions as well as the interesting answers provided by the panel. In addition, I will provide a brief overview of other insights I gained at the conference. (Now, if I could just get over this jet lag!)

Bookmark and Share

Posted in ADAS, Data Sources, Geospatial, Intermap, Mapping, Mike Dobson, Navteq, routing and navigation, TeleAtlas


(comments are closed).