There are no standards for defining APC accuracy in the U.S. transit industry though progress is being made through organizations like APTA and some industry stakeholders.
Passenger data improves planning, increases efficiency and positively affects service. But, it starts with the right technology. Choosing the low-cost passenger counting system is not always the best solution. Better results begin with more competent technology as well as an established standard for measuring accuracy.
Establishing the Standard
Automatic passenger counting (APC) systems are designed to gather accurate ridership data so that agencies can provide optimal planning for better service. Using ridership and on-time performance information can help in the evaluation of routes, which enables planners to make educated decisions about service schedules.
Since modern APC systems can provide important ridership data in a reporting interface by the next service day, the service vehicles can be automatically matched through the evaluation software with the schedule data.
This data is significant for agencies because in order for them to be eligible for the Federal Transit Administration (FTA) program funds, they must meet a set of data requirements. According to the FTA, transit agencies are required to "report
to NTD (National Transit Database) annual data on unlinked passenger trips (UPT) and passenger miles traveled (PMT) for each mode and service type (purchased versus operated). They must report a 100 percent count of each quantity from [bus
and rail] if it is available and reliable. If a reliable count of 100 percent is not available for a quantity, they must estimate it
through random sampling and the obtained estimate must meet the minimum confidence level of 95 percent with a precision
level of ± 10 percent."
The problem is there are no standards for defining APC accuracy in the U.S. transit industry — as of yet. Progress is being made through organizations like the American Public Transportation Association (APTA) and some industry stakeholders. But, until the requirements are defined, any APC system provider can claim that their product provides a high accuracy rate.
What to Look For in RFP
When releasing an request for proposal (RFP) for an APC system, some simple questions are in order. The most important of these relates to the data itself.
- By what standard is the APC accuracy measured?
- What is the credibility of the count information provided?
- Has the system been proven by a defined standard?
How count accuracy is measured is important. Very often, RFP documents require a parameter of greater than 95 percent
accuracy. This appears to be a reasonable measurement — on paper. The challenge comes when vendors promise these agencies that they can meet high accuracy rates without providing substantial evidence of how they are measured.
For example, an excerpt from a recent RFP reads "An automatic passenger counting system shall be installed to record boardings and alightings at stations and stops. The system shall be rated for continuous duty and provide an accuracy rate of 96 percent."
Indicating the sensor system "shall have a minimum 96 percent counting accuracy" is subjective. Such broad statements in the RFP have been known to later generate countless discussions between the supplier and agency about accuracy before getting final approval of the system.
In Europe, the VDV 457 standard is used on all APC installations. The standard was established by the VDV, Verband Deutscher Verkehrsunternehmen e.V., an industry organization similar to APTA. The VDV requires new APC projects to get final acceptance for ridership data accuracy from a certified third-party auditor.
The requirements include a tested formula for generating valid passenger count data.
Definition of accurate passenger accumulated count data:
erel=(|Pa-Pm\)/Pm • 100% ≤ 2%
- Pa = automated counted passenger (each separate for in and out)
- Pm = manual counted passenger (each separate for in and out)
- sample size is min. 1,000 passengers (Pm>1,000)
Based on this formula, the accumulated count of alighting passengers shall be within 2 percent for each 1,000 consecutive boarding and alighting passengers, and the accumulated count of boarding passengers shall be within 2 percent for each 1,000 consecutive boarding and alighting passengers as outlined in the diagram.
In addition, to provide accurate stop-by-stop count data, 85 percent of all door cycles with the boarding and alighting counts shall be exact when compared to actual. For 90 percent of the stops, the counts shall be within one passenger of actual. For 97 percent of the stops, the counts shall be within two passengers of actual. This includes stops for which there was no observed boarding or alighting activity. Counts from multiple door openings near a stop shall be correlated to that stop.
More and more agencies are asking for real time APC data which comes with the requirement of higher accuracy counts on the vehicle. The challenge is that outdated technology is still promoted by some vendors as a reasonable, low-cost solution. These systems count on post-processed data that has been manipulated in order to meet RFP requirements. Using older technology and promising higher results does not yield truly accurate results. Today’s transit environment demands better tools for the most accurate results.
Init, a global provider of ITS technologies, uses the VDV standard with all its projects worldwide. This standard demands the highest accuracy rates based on the actual raw data.
In the images above, there is an example evaluation of Init's standard accuracy calculation based on the actual raw data.
The calculations measure if the data meets the standard of less than 2 percent error on 1,000 boarding and alighting passengers. In addition, stop accuacy is also measured for the passenger load to calculate passenger miles traveled (PMT).
Sensor Technology Clarified
Sensor technology is constantly evolving. In the past, a simple infrared beam in the side of a door, or a one-dimensional active sensor was sufficient. Today however, the technology has advanced to video-based or time of
flight (TOF) technologies which not only provide smarter and more accurate results, but also deliver higher accuracy
on raw passenger data.
Init uses the IRMA Matrix sensor which is based on TOF technology and makes use of an innovative and entirely new detection method. Light pulses are sent out in quick succession using the invisible IR range. These pulses are reflected by objects and detected by the sensor. The distance to the object is then calculated by taking the period of time between transmission and reception of the light pulse. In this manner, the system generates a three-dimensional image of the door space, so that individual persons may be detected even in tightly packed crowds.
Unlike conventional sensors, the Matrix sensor does not need the door contact signal which has often been a source of malfunction in the past. The IRMA Matrix is installed overhead and provides the highest verifiable accuracy data on raw passenger data. It can be installed as an external or an internal flush mount above the vehicle's entry/exit door.
Quality Assurance
Init uses the IRMA Test Ride software to demonstrate accuracy. With this tool, agencies can easily validate the installed components. We calibrate, verify and certify through a step-by-step process.
Once verified through the Test Ride process, an APC certification is issued based on the VDV standard. Afterwards, agencies can easily check the quality of the APC system each day without manual counting using the reporting system to
evaluate the APC quality indicator.
The quality check for boarding and aligthing passengers is around 3 percent for the Matrix sensor. That should be a very
good indicator of whether or not the APC system has any issues.
Customer Value
The value of having an automated ridership counting process is to have as much detailed information on your
ridership by trip, by stop, by route; etc. for any given time period. Being able to aggregate the data will allow you to perform better reporting and more efficiently manage your resources. This is why it is important to make sure the count
accuracy is measured and verified so that you can produce tailored services and provide reliable reports to NTD.
Init has a clear understanding of the criteria in regard to precise passenger counting data. This is demonstrated by our commitment to follow the VDV standard of accuracy testing for all our APC projects. Precise accuracy will allow you to obtain accurate data for the long-term integrity of your system. The best course of action is to do the homework before you buy, and consult with someone who is experienced and knowledgeable prior to the RFP release.
Andreas Rakebrandt currently serves as Init’s director of Business Development for APC technology.
