Tests by engineers at the University of Minnesota suggest that only a presently expensive form of satellite based GPS called NDGPS is accurate enough for tolling. Only the highest quality in-vehicle receivers costing $3,000 apiece and improved digital maps will do the job... MORE
Tests by engineers at the University of Minnesota suggest that only a presently expensive form of satellite based GPS called NDGPS is accurate enough for tolling. Only the highest quality in-vehicle receivers costing $3,000 apiece and improved digital maps will do the job.
The N of NDGPS stands for a Nationwide US system and the D for differential. DGPS receives corrective terrestrial signals that refine the rougher satellite based calculations of straight GPS. DGPS with very expensive dual frequency receivers is widely used by surveyors so in that static situation it is capable of accuracy within inches.
GPS as commonly used by in-vehicle mapping devices or for other general location finding such as GM's OnStar is deemed inadequate for tolling, since it only guarantees accuracy to within 13m (43ft) with 95% confidence. For tolling there is a need to locate a vehicle in a specific roadway since a system must distinguish between a tollroad and a free frontage road or just a local road maybe 5m (16ft) away.
5 percent of interstate highways in US urban areas have another road within 15m (50ft) and 20% have another road within 30m (100ft). Ideally too, given the interest in HOT lanes, a GPS system should be able to locate a vehicle in a specific lane of a roadway, since a toll lane may be located right alongside a free lane. In this case more than an order of magnitude better accuracy is needed than straight GPS at least for lateral measurements across the road.
NDGPS is planned to provide a corrective signal providing accuracy to 1m (40in) over the whole of the contiguous 48 US states by the end of 2005 with cheap receivers. Minnesota is already covered so the tests done there were something of a preview of coverage that should be available nationwide mid-decade.
The U Minnesota vehicle tests of NDGPS used a calibrated digital camera mounted directly under the DGPS antenna, plus square tiles placed in the roadway at strategic points with which the camera could establish a baseline or "ground truth" against which to test the DGPS system. Repeated runs were done at different speeds to test accuracy.
Tested were a Trimble AgGPS-132, a high end $5k unit, an $850 CSI GBX-12R, and a $500 JRC DGPS-212. Tests were run over nearly 1,000km (620mi) of interstates and other major highways around the Twin Cities metro area at travel speeds that averaged 74km/hr (46mph).
Sure enough the fanciest unit had the least error. The expensive Trimble stood out!
The 3-digit $ units don’t hack it. The 4-digit $ one does. Big difference in precision.
997 out of a thousand measurements would be within 0.2m (8in) whereas with the cheaper receivers such a proportion of measurements would only be within 8.3m (27ft) and 11m (36ft).
The researchers do comment that a slimmed down version of the champ Trimble is available with the same accuracy in this application, and that prices should come down substantially over the next several years.
They say there are three reasons that money buys a better result: (1) processing of carrier and code phase signal by the AgGPS-132 vs code phase only by the cheapies (2) a better antenna on the AgGPS-132 (3) better receiver software.
Most digital maps presently in use aren’t adequate for supporting NDGPS tolling the report says, but adequate maps are available or can be made.
They suggest the possibility for successful use of GPS for tolling is there, but needs more work and larger field trials before the right technology can be selected. (Researchers: Pi-Ming Cheng, Max Donath, Xiaobin Ma, Shashi Shekhar. Contact pmcheng@me.umn.edu)
COMMENT: The Europeans don’t seem to be going for anything like the accuracy the U Minn researchers suggest is needed for the US. The Germans and Swiss talk of systems accurate to within “several meters” apparently not sufficient to get accurate lane by lane placement.
But why not a combination? Why not reliance on satellite in the sparsely settled areas where sub-meter accuracy is not needed and where DSRC infrastrcture would be expensive to install and DSRC in the urban areas where there’s a ton of infrastructure anyway and you need the far greater positional accuracy that is easily accomplished from a few meters overhead. Why stretch the satellite technology so much. There’s no need.
Here we’ve got perfectionist engineers fixated on one technology (GPS), and that fixation makes the job far more difficult and expensive than it need be if the full range of technologies is considered. TRNews 2003-09-10
Add table.
