|
|

Brief APC Tutorial
|
Back
to Forum Page
APC Terminology
| Term |
Definition |
| Active IR |
Near-IR light-beam sensors that transmit light
continuously, or in pulses, and then receive that light. See Through-beam
IR, Retro-reflective
IR, and Diffuse-reflective
IR. |
| Algorithm |
Data processing technique for turning raw sensor
data into passenger counts. See Finite-state
machine, Neural net,
and Image processing. |
| APC |
Automatic Passenger Counting |
| AVL |
Automatic Vehicle Location |
| AVLC |
Automatic Vehicle Location and Control |
| AVL/CAD |
Automatic Vehicle Location/Computer-Aided Dispatch |
| Contact APC |
APC systems that such contact-type sensors,
such as treadle switches, smart mats or switch matrices. |
| Crush Load |
Passengers entering an already-full vehicle
crowd into the door area, producing a crush load. |
| Data Retrieval |
Retrieving the APC data from the vehicle after
it is collected. See Offloading |
| Dead Reckoning |
A navigation technique that uses speed and heading
to determine location. Sometimes called, "ded reckoning," short for, "deduced
reckoning." |
| Diffuse-reflective
IR |
Near-IR light-beam sensors which project a beam
of light that reflects off an object, such as a passenger. See Active
IR. |
| Event-based Data |
APC data in which every passenger movement (on
or off) is time-stamped or otherwise identified. |
| Finite-state Machine |
Algorithm in which the raw sensor data causes
the computer program to step through a series of pre-determined states
to determine passenger counts. |
| GIS |
Geographical Information System |
| GPS |
Global Positioning System |
| Host System |
On-board electronic system to which APC equipment
is connected, such as AVL, AVLC,
MDT,
etc. |
| Image Processing |
Algorithm which takes images from a vision system
and processes them, using special electronic hardware or a microprocessor,
into passenger counts. |
| Infrared (IR) |
Can mean either near-IR - wavelength just slightly
longer than visible light - or thermal IR - wavelengths between 5 and 10
µm. |
| Load |
Number of passengers on a vehicle at a particular
moment. |
| Machine Vision |
See Vision system. |
| MDT |
Mobile Data Terminal |
| Neural Net |
Algorithm in which data filters through a probability
matrix to determine passenger counts. Can be implemented using special
electronic hardware, or programmed into a microprocessor. |
| Non-contact APC |
APC systems that use non-contact sensors, such
as active IR, passive IR, vision or ultrasonic. |
| Offloading |
The process of retrieving data from a vehicle.
This process may be manual, or automatic, using optical or RF modems, or
a direct data link through the vehicle radio system. |
| On-board processing |
Raw APC data is processed into passenger counts
by a microprocessor on the vehicle. |
| Passive IR |
Thermal-IR sensor that detects changes in temperature,
such as moving passengers. Typically located overhead, these are similar
to motion detector security devices. |
| Post-processing |
Raw APC data is recorded on-board the vehicle
and processed into passenger counts after data retrieval |
| Real-time APC |
APC data is available at the moment it is collected. |
| Retro-reflective
IR |
Near-IR light-beam sensors with transmitters
and receivers on the same side of the door area. Horizontal light beam
is reflected off a retro-reflector on the opposite side of the door area.
See Active IR. |
| RF |
Radio Frequency |
| Smart Mats |
Rubber mats with embedded optical fibers that
sense deflection, and therefore, foot placement. |
| Stop-level Data |
APC data which reports the on and off passenger
counts at each stop, and may include the load at the end of the stop. |
| Stops Database |
Database of stops that is related to geographical
and route/schedule information. |
| Switch Matrices |
Rubber mats with a array of many embedded switch
elements |
| Tare |
Passenger load at a particular reference point,
such as the end of a trip. Used for establishing a reference load for real-time
APC systems. |
| Time-based Data |
Data in which passenger counts are accumulated
during discrete time intervals. No stop or location information is recorded. |
| Through-beam IR |
Near-IR light-beam sensors with transmitters
and receivers on opposite sides of the door area. See Active
IR. |
| Treadle Switches |
Rubber mats containing embedded switch elements. |
| Ultrasonic Sensor |
Acoustic sensor which senses passengers by transmitting
and receiving ultrasonic pulses. |
| Ultrasound |
Sound frequencies above the range of human hearing,
typically 30 to 100 kHz. |
| Vision
System |
System in which an electronic camera
takes pictures of the door area to determine passenger counts. |
Why APC?
Need for APC is determined by requirements for scheduling & planning,
reporting, and fleet operation. Here are some typical uses:
· Total counts for fare verification
· Collecting ridership data for scheduling
and planning
· Schedule adherence
· Real-time load for operational control
Data Collection
& Retrieval
The APC data has to go somewhere. For real-time APC, it is passed to
a host system where it is transmitted to a central location over the vehicle
communications link. In many APC systems, the "store and forward" type,
data is stored on board the vehicle for later data retrieval. It can be
stored in digital memory, in a removable data card, or on a diskette.
Data retrieval can be manual or automatic. Manual data retrieval, sometimes
called "sneaker net", requires that someone visit the bus and remove something
(a data card or diskette), or plug something into the APC equipment, like
a notebook computer. By equipping the vehicle with an RF or optical modem,
the APC system can automatically connect with an external computer and
download the APC data without human intervention.
Location Information
To make stop-level data meaningful, the data has to have location information.
Otherwise, how would you know which stop is which? Unless the APC equipment
is connected to a host system with location capability, APC needs another
component. This other component could be an RF or optical signpost receiver,
a GPS receiver, or perhaps dead reckoning equipment.
Signpost systems use strategically placed RF or optical transmitters
to send a code to the passing vehicles. The codes are stored along with
the other APC data to determine the vehicle location through post-processing.
Of course, this requires that the vehicle have an on-board RF or optical
signpost receiver.
Alternately, each vehicle is equipped with a GPS receiver. Position
is determined by receiving signals from 3 or more satellites. The receiver
performs the complex calculations, and the APC system stores with information
along with the APC data. This can be more complicated than it sounds, but
doesn't need the signpost infrastructure.
Dead reckoning can be used by itself, or with signpost or GPS equipment,
to provide location information. The vehicle's odometer signal, plus signals
from an electronic compass or other heading indicator (like a gyroscope)
are processed to determine the vehicle's velocity (speed and direction).
This information is stored along with the APC data to determine vehicle
location through post-processing.
Accuracy
This is complex subject, because it involves statistics. APC accuracy
can range from around 80% to better than 95%, depending on the type of
APC equipment, the configuration of the vehicle door area, the types of
loads, and the behavior of passengers. Accuracy is always better when loads
are lower, and worse when loads are high.
Generally, accuracy is measured by comparing APC data to manually-collected
data. Manual data, which is hard to collect in heavy load situations, may
only be 90 to 95% accurate, making it difficult to evaluate accurate APC
systems. Video taping, followed by careful analysis in the comfort of the
office, is the most reliable method of evaluating accuracy.
Sometimes, the accuracy specified for an APC system is stated as absolute
- "The APC system must be 95% accurate." This is not a meaningful statement,
because it does not specify a confidence level for this measurement - a
common trap in statistics. Put simply, any system will encounter rare conditions
that will cause it to make mistakes. An APC system may be really accurate,
but every once in a while, the accuracy will be terrible, and the data
should be thrown out. If it desired that the system be 95% accurate, but
it is acceptable to throw out 5% of the data (usually on a trip basis),
a better way to specify the accuracy is, "the APC system must be 95% accurate,
95% of the time".
Specifying
an
APC System
Here are some things to think about before specifying an APC system.
1. What is the purpose of the APC system?
As obvious as this sounds, for the APC system to meet the customer's
needs, the customer must know the problem to be solved. Since APC is automatic,
lots of data will be available, but to what end?
2. Will the APC equipment be connected to a host
system?
If so, interface and data format requirements must be carefully thought
out.
3. Is stop-level data required?
In many cases, stop-level data is not required and the APC equipment
can be greatly simplified. Stop-level data also implies the need for location
information, creation of certain databases for post-processing, and, in
general, greater complexity.
4. What accuracy do you really need?
Many APC systems are over-specified, resulting in higher cost and complexity
than is really necessary. For example, real-time systems, such as AVLC,
that need APC load information only, can generally get by with lower accuracy,
providing the system has a way to tare the load at strategic points.
5. How many APC-equipped vehicles are necessary?
For AVLC systems, where real-time load is desired, each bus needs to
be equipped with APC. However, where the APC data is to be used to scheduling
and planning purposes, a rule of thumb is to equip 10% of the fleet.
6. How will the data be retrieved?
Manual data retrieval may be fine for smaller fleets, but consider
automatic data retrieval for larger systems.
7. How will the data be processed?
APC equipment is automatic! That means data is being collected whenever
the APC-equipped vehicles are in service. This can result in massive "data
glut." Consider the amount of data to be processed, the equipment required
to process it, and the programming effort required to create or support
the system. Some APC implementations have failed because the data processing
requirements were greatly underestimated.
8. How many people will be needed to retrieve,
process the data and generate reports?
While it's true that, with an APC system, you won't be sending many
people out to collect data in the field, the number of people required
to manage the system, retrieve and process the data, and generate reports
may be significant, depending on their skills and the data processing software.
9. Consider all the costs
APC systems need not be excessively costly, compared to other similar
investments in equipment and personnel. In fact, a properly specified and
operated APC system will produce a net cost savings through lower personnel
costs and increased operating efficiency. However, consider all the costs
involved, including acquisition, installation and life-cycle costs of the
equipment, plus system and data management.
Conclusion
Those transit authorities that have installed and used APC systems generally
feel that they can't get along without them. However, like any new method,
specifying, installing and learning to use an APC system takes money, time
and effort. We hope this little tutorial has at least helped to illuminate
some of the darker corners of the APC world.
Back to Top
Updated 08
Feb 2002
|