Source: TOLLROADSnews 2008-10-28
The American offices of Kapsch are reporting "extraordinary success" in a performance evaluation of their 5.9GHz electronic toll system in trials on E470 in Colorado recently. In a press statement they say a nationally known, independent R&D laboratory evaluated the system and determined that it collected "100%" of more than 10,500 DSRC sample passes.
The transponder/reader or DSRC system was tested by a fleet of 27 vehicles which made approximately
11,000 passes under the Kapsch readers. Comparing the transponder reads with a count of vehicle passes using a separate GPS system the independent laboratory concluded that the system obtained 100% accuracy.
On request Kapsch provided us a copy of the report of the testing by the independent laboratory minus the cover page, and with the name of the laboratory blacked out. They said the laboratory demanded a non-disclosure agreement.
We discovered independently of Kapsch that the laboratory was the Southwest Research Institute (SwRI) in San Antonio Texas, so we'll refer to the report as the SwRI report.
E470 Public Highway Authority allowed them to mount their readers, vehicle detection and classification lasers, cameras, and lights on a beam over one set of open road toll lanes at the Parker Road or Toll Plaza A on the far southern end of tollroad. Their equipment was set up without interference to the 915MHz Title 21 toll system that takes tolls routinely there. The plaza services about 17k vehicles/day.
SwRI says the tests were also of the Kapsch laser vehicle detection and classification (LVADC) compared to loop-based VDAC, a Kapsch automatic license plate reader as well as the 5.9GHz reader/transponder 
communications. Test specifications are to be published separately.
Tests were conducted weekdays over two weeks Aug 25-29 and Sept 1-5, 8am to 6pm. During the first week the test drivers drove only in lanes, but in the second week they did straddles of lanes, including the shoulder lane.
Kapsch personnel mounted the transponders in the test vehicles. They used mounting brackets with two 
different angles for differently angled windshields.
In some tests three transponders were fitted to the one vehicle to simulate closely spaced vehicles.
10,000 passes were the target sample size, SwRI says because Kapsch believed they could meet a 99.9% accuracy.
SwRI mention in several places in their report that the test drivers were under strict instructions to drive safely and a proper distance from other vehicles. For example: "The safety and security of the drivers and equipment... was given the highest priority. Drivers were instructed to stop if hazardous weather conditions occurred."
SwRI indeed reports 100% read results for all 27 drivers in both single tag and triple tag (called "Over Equipped") tests for a total of around 11,000 passes.
COMMENT: 100% is an extraordinary read rate! Perfection. We've heard claims of 99.95% and above in tests, but never 100%, at least not in test samples this large.
97 or 98% is considered good for transponders in actual use, and some do only 95 to 96%.
The SwRI test had the advantage of transponders mounted in cars by Kapsch staff, versus the norm of customer mounted transponders in the real world of electronic tolling, at least in the US. Customers make mistakes in mounting transponders. Some end up just holding them to the windshield and storing them in the glove box the rest of the time.
Also on real roads a proportion of drivers don't obey the safety rules of the kind SwRI laid down for their test drivers. Some tail-gate and get close to high trailers whose metallic mass can scramble the RF signals.
Real drivers keep driving in nasty weather where SwRI testers called it a day.
How far the three tag per vehicle set-up simulates close-spaced vehicles, we're unsure.
5.9GHz transponders are designed to be factory installed whether made by Kapsch or any other manufacturer so all brands would have a mounting advantage over present models.
All that said 100% accuracy for transponder reads in nearly 11k passes can't be beat.
VDAC detection accuracy of 99.74%, classification 88.62%
SwRI reports the Kapsch laser vehicle detection and classification system (LVDAC) in the same 11k+ test drives had a 99.74% rate. Only 31 out of 11,912 vehicle passes were not detected.
The detection rate plays into all the other operations since it is usually a 'trigger' for the reader, for classification, and the camera. Tou don't read if you don't first detect.
Loops, apparently Idris loops, in use by E470 had a tad better detection rate, SwRI reports - 99.76% But the difference, 99.76 and 99.74 is probably not statistically meaningful.
SwRI did two sets of classification tests of the Kapsch LVDAC units. The more stringent test required separation of Class 2 and 3 vehicles, the less stringent accepted them grouped as one class. Results were 88.62% accuracy and 96.86%. (FHWA F-series classification)
The loops do better in FHWA-F classification than the LVDAC - 92.32% and 98.07%. This is not surprising since the vehicle classes are based on axle count, and loops located where they are in the roadway seem inherently more suited to counting axles than any kind of overhead equipment which has real problems getting a useful angle of view of the wheel set.
LVDAC are usually used - as in 407ETR Toronto - for classing based on vehicle dimensions and size and distinguishing cars, straightbody trucks, and articulated or combination vehicles.
License plate reads 93.84% rear, 91.28% frontal
The Kapsch automatic optical character recognition expressed as a percentage correct of humanly readable plates with roughly the same 10k+ passes got results of 91.26% accuracy for frontal reads and 93.84% for rear camera reads.
SwRI make the unexceptionable assessment that results in use of the Kapsch gear should be similar in similar traffic conditions, on similar roads, using similar equipment, under similar environmental conditions.
BACKGROUND: Kapsch is a leading supplier of 5.8GHz European standard transponder/reader systems in Europe, Australia and South America. Headquartered in Vienna Austria it is a public company traded on the Vienna stock exchange and has grown in part by acquiring leading Swedish and Germany toll systems companies.
In June Kapsch bought up the Mobility Solutions division of TechnoCom Corp of Encino Califoprnia which has been a leader in 5.9GHz communications for the federal government supported Vehicle Infrastructure Integration (VII) program aimed at using the most modern technology for vehicle-to-vehicle and vehicle-roadside for a whole range of safety, convenience and commercial applications, including tolling.
Kapsch reports it has 200 projects working in over 30 countries with over 13 million transponders and other on-board units and over 11,000 equipped lanes. This includes parking payment lanes, access control lanes, special telematics lanes as well as toll lanes.
Kapsch TrafficCom US Corp, the American operation now has establishments in Sterling VA near Washington Dulles Airport and in California. They seem to be making a major push for business and represent a strong competitor for Raytheon, TransCore, ETC/Autostrade, Telvent/Caseta and other established US groups.
TOLLROADSnews 2008-10-28"
- First 5.9 GHz DSRC VII Network and Toll System Established on New York City Streets and Highways
- First 5.9 GHz DSRC VII Network and Toll System Established on New York City Streets and Highways: Kapsch TrafficCom United States Installs 40 Units
- Kapsch deploys wireless roadside system in New York
- Big Success with the Performance Evaluation of the New 5.9 GHz Tolling Technology from Kapsch TrafficCom at the Trial Facility in Denver, Colorado.
- Kapsch deploys wireless roadside system in New York
Keine Kommentare:
Kommentar veröffentlichen