Monday, December 17, 2012

Xerox Admits 5% Error Rate

Baltimore City's Speed camera Vendor, Xerox Corporation, has admitted that at five of the city's speed camera locations over five percent of citations issued were in fact the result of errors.

In the two page report released by Xerox, the company stated that they had examined a total of 7000 citations and concluded that at five locations 5.2% of citations were errors that slipped through the review process and were issued.
Report on Camera Errors By Xerox, see PDF version here
"In a limited number of cased, radar effects can occur.  Radar effects are caused by reflection, refraction and absorption"... 
"Based on review of the images alone, we identified five (5) locations that demonstrated higher incidents of radar effects than is typical and need further investigation to determine the root cause."... 
"As part of the review, Xerox examined over seven thousand issued citations for radar effects that may have resulted in erroneous speed recordings.  With respect to seventy eight percent of the eighty three (83) locations, the review showed incidence of radar effects that were not identified in processing of approximately one-half of one percent (.5%).  Five (5) of the eighty-three(83) sites revealed some environmental issues causing an unusually high rate of occurrence of radar effects, typically attributed to the presence of high profile vehicles in the field of view.  A more extensive review of these locations showed incidence of radar effects that were not identified in processing of 5.2 percent (5.2%). The five locations are
* 1400 Block of Cold Spring Lane E/B
* 1700 Block of Cold Spring Lane W/B
* 400 Block of Franklin Street W/B
* 500 Block of Cold Spring Lane E/B
* 1300 Block of Cold Spring Lane E/B

A copy of this report was provided to by Baltimore City officials.

Xerox entered the speed camera business when it acquired photo enforcement company ACS State and Local Solutions.  They are also the speed camera vendor for Montgomery County, the SHA, Baltimore County, Howard County, Rockville, Gaithersburg, Takoma Park, Chevy Chase, Bowie, and Frederick.

The company claimed that the error rate at other locations was 0.5%.  However taking even this small percentage at face value would mean thousands of false accusations were being made annually in a program the size of Baltimore's, which issued about half a million citations last year.  We also note that .5% is one in 200, so it would seem that in order to confirm that ALL 83 sites have rates no higher than .5% would require reviewing at least 16600 citations, so 7000 (the majority of which are at the 5 sites listed above) would not have been sufficient to prove statistically that all other cameras had error rates no higher than 0.5%.

The report came after obtained documents proving that Xerox and city officials had knowledge of speed measurement errors at one location in February and July of this year, after complaints by companies about erroneous citations issued to trucks and delivery vehicles, and after an in-depth investigation by the Baltimore Sun which discovered that timestamped images and videos from violations at several cameras proved that some citations showed that vehicles were not in fact speeding.  In the most egregious incident, the Sun discovered that a vehicle was cited for speeding that was not even moving.

The admission by a photo enforcement company that 1 in 20 citations were the result of errors is almost unprecedented in the industry.  When claims of erroneous citations were issued in other parts of Maryland from cameras run by Optotraffic in jurisdictions such as Forest Heights, College Park, and Brentwood, both the contractor and local governments dug in their heels and claimed that citations images could not be used to disprove speeding.  Even after the town of Cheverly released documents proving that Optotraffic's cameras did produce errors, the company refused to admit to any errors.  Optotraffic even went so far as to remove the 'delta time between images' and to reduce the timestamp precision, in order to ensure that motorists would be unable to use these to prove speed measurement errors.  In the case of the town of Brentwood, the results of a supposed investigation by Optotraffic into alleged errors was never made public, and Brentwood has yet to provide any documents in response to two public information act requests made by this website, despite the passage of two years and numerous attempts to obtain them.  The town of Forest Heights refused to provide records in several instances as well.

Other jurisdictions's speed camera programs (including some run by Xerox)do not have detailed timestamps to fractions of a second on citation images, nor do they have videos, making it possible that errors of the sort seen in Baltimore could occur with no evidence that could be used to conclusively prove it.

The Xerox report maintained that "The Speed enforcement systems have been operated by the City consistent with industry standards and in compliance with applicable law", and that "the review demonstrates that the speed enforcement systems are reliable and accurate." in the report.  "There have been some very isolated incidents of human error. We've seen that in every program that every vendor operates in the United States.",  Stated Xerox VP of operations Allen Schutt in video interview recorded by the Baltimore Sun.

Additional Coverage: