VEHICLE INTELLIGENCE & TRANSPORTATION ANALYSIS LABORATORY
    University of California, Santa Barbara

ITS Datum

 
Most of this page is more than a year old.  Resolution has been achieved on many of the "outstanding dilemmas" below.  Watch for an update soon.  The section on Rubberstreeting (end) has been added recently.


The ITS Datum, proposed by Oak Ridge National Laboratory (ORNL), is a planned network of monuments across the United States, that could potentially serve as

  1. a positional reference for ITS applications corresponding points in commercial databases are identified, and rubbersheeted on the fly to the more accurate Datum coordinates;
  2. a uniquely identified set of reference markers (nodes and links), relative to which locations may be expressed, and distances measured.
In the long term a Datum may evolve into an accurate standard nationwide street network.  Commercial vendors would add value by attaching attributes and application software, such as navigation, traffic management and address geocoding.

A prototype Datum has already been developed, based on the National Highway Planning Network (NHPN) database.  It consists of about 50,000 points across the country.  The illustration at right shows the county of Santa Barbara, the NHPN in gold, and the prototype Datum in black.  Points in this prototype are defined to +/- 80 metres.  Clearly this is not sufficient for many applications.  The network will have to be densified before it can serve a useful purpose in ITS.  Prior to an expensive national scale densification effort, questions need to be answered on

If a national infrastructure such as this could serve similar needs in application areas other than ITS (e.g. GIS-T), then those user requirements should be considered in the design process.

To Be Resolved

It is clear that the Datum is a necessity, to improve the accuracy of location messaging.  We have documented gross disagreements between street network databases (example), and tests of the Cross Streets Profile show that messaging standardization alone is not sufficient for unambiguous communication.  Future changes, whether geoid standards or tectonic movements, demand a flexible referencing structure.  The concept and methods pertaining to the ITS Datum are relatively well understood.  What is yet unclear is the performance requirement: what level of accuracy it is required to achieve, and at what cost.  Consultations with user groups are underway to gain a better understanding of this.

Point density is one of the principal determinants of cost.  Should Datum points be located at major highway intersections only, every major street intersection, every residential street intersection, every driveway?  The figure below shows three different ways of capturing a freeway intersection.  The first employed in the 50,000 point national prototype uses one point to represent the entire interchange; the second considers the intersections of carriageways; the third looks at each decision point (based on drive-on-the-right; British readers consider the other 8 points in the intersection).

Similarly there are different ways of capturing city streets.  One model considers the intersection of centerlines; the other surveys a point on each incident link, e.g. at the STOP line in the right hand lane.  Note that the first can be derived from the second in most cases.

On the matter of resolution and accuracy, some argue for an engineering-grade survey of Datum points, i.e. correct to a few centimetres.  An obvious difficulty is that centrelines and kerblines are themselves nebulous targets, that wander with each re-lining and re-surfacing of a street.  Does a Datum point then require a physical stake and brass marker in the ground?  An associated question is how the point is to be surveyed by kinematic GPS using a van traveling at highway speed, or a static survey.  The costs differ significantly.

Then there are placement rules.  A standard would require that each authority charged with establishing the Datum must use a prescribed procedure for determining appropriate points.  Exactly where does one place a point at a freeway exit?  Some suggestions:

Gore point and off-road point
Gore point wandering: this one moved 9.5 metres when the right shoulder line was repainted 0.6 metres left of its earlier position
Decision point showing zone of uncertainty

Incremental Development?

Clearly there is uncertainty in all of the above approaches, due to (a) determining the point to be fixed, (b) method of survey, (c) variability in the location of the point over time.  It seems inevitable that at least in the short term, a Datum point can be fixed to no better than 510 metres on freeways, and 13 metres on urban streets.

A longer term approach is to plan incrementally.  Consider the 50,000 point Datum, already created, as Phase I.  In Phase II (say to the year 2005):

Phase III kicks in when it is deemed necessary and practical to improve accuracy to less than a metre.  A brass marker may be established for each cluster (the red star in the illustration, not necessarily in a paved area) and local Datum points surveyed relative to it, with whatever precision is appropriate.  Futuristic technologies, e.g. beacons triggered by vehicle sensors/identifiers, may transmit accurate and current data on the intersection or neighbourhood, to passing vehicles.

Database Structure

We have used the terms "nodes" and "links" above in reference to the ITS Datum.  This terminology has caused some confusion in the past, because some readers think it implies that every node and link in the physical street network has to be re-surveyed to great precision.  It should be obvious from the discussion above that this is not the case.

There is no immediate need for the Datum to be a topologically structured network (this would be a longer term requirement if the Datum were to evolve into a skeletal national street network).  As new points and lower-level objects are inserted, topological enforcement would require reassignment of IDs.  We are examining ways to preserve topological integrity while not discarding IDs.  Further details in the proceedings of the technical issues breakout group from the first workshop on public sector user requirements.
 

Rubberstreeting

January 1999.  VITAL has developed an algorithm for geometric adjustment of discrete topological fields (i.e. street networks) that can be applied to correcting maps in real time using the ITS Datum.
 

Larger image
Residential neighborhood example

Before adjustment: two maps show positional discrepancies up to 100m. A vehicle driving down the blue centerline (the true alignment of the road) appears to be in someone's backyard relative to the pink centerline (from vendor E).


Larger image
After E is rubber-streeted: the maps are almost in complete agreement, and the vehicle is correctly positioned on both the pink and blue centerlines.

 

Larger image
Freeway interchange example

Before adjustment: a westbound vehicle (blue dot in the center of the red circles) exits to the southbound ramp, as can be seen in the blue database, but with respect to the pink centerline, it appears to be still on its original course in the westbound lane (green dot).

Larger image
(coming soon)
After E is rubber-streeted ... the problem is not entirely solved, in fact parts of the interchange are transmogrified.  The problem is that database E misrepresents not just the geometric alignment but also the topological relationships in the interchange.  The Datum has to be designed to overcome this, by means of a subsystem of topological specification and ramp identification, in addition to just authoritative coordinates.

Finally ...

The ITS Datum promises solutions in ITS, but many questions remain unanswered with regard to its design and deployment.  We are in the process of consulting public sector users to gauge the performance requirements of the Datum.  Further details under Meetings.

Last update 1999-04-15


Research  Home Page