Vehicle Detection Systems

There are several types of vehicle detection systems (VDS). The earliest of these is the inductive loop, which is a wire looped under the road surface. Some systems use radar mounted on the side of the road. Video detection uses a camera and computer vision software. Collectively, these systems are simply called detectors.

Configuration

To create a detector, first select the r_node at the proper location. Then select the Detectors tab for that r_node. Enter the detector Name and press the Create button.

After selecting a detector in the r_node detector table, its properties, such as lane type and lane # can be changed. Lanes are numbered from right-to-left, starting with the right lane as 1. A label will be created from this information, including abbreviations of the roads associated with the r_node.

The field length of a detector determines how density and speed are estimated from counts and occupancy. It is in units of feet.

If a detector is no longer used, it can be marked abandoned.

Lane Type

Transfer

It is possible to move a detector to another r_node. Select the target r_node and enter the detector Name. The current label for that detector will appear on the right. To move it to the current r_node, press the Transfer button.

Traffic Data

The preferred method of collecting traffic data, supported by newer detectors, is to log every vehicle that passes by the detection area. Older detectors use fixed time intervals and put data into bins.

Every 30 seconds, an XML file is generated containing the most recent sample data from all online detectors. The file is called det_sample.xml.gz, and it is written to the XML output directory.

Detector Protocols

IRIS supports several different protocols for communicating with vehicle detection systems. The protocol used depends on the comm link of the controller to which it is assigned. The following table summarizes features of each protocol.

For protocols which allow the binning intereval to be adjusted, it will be set to the poll period of the comm link.

Auto Fail

Traffic data is continuously checked for five common failure conditions. When one of these first occurs and every hour that it persists, an event is logged in the detector_event database table. The detector_auto_fail_view can be used to check recent events.

If the detector_auto_fail_enable system attribute is true, the auto fail flag for each detector will be set and cleared automatically whenever these conditions change.

No Hits

This failure condition occurs if no vehicles are counted for a duration determined by the lane type. It clears immediately when a vehicle is counted.

Chatter

If a detector reports an unreasonably high count of 38 vehicles or more in a 30 second period, this condition will be triggered. It will be cleared if 24 hours pass with all counts below that threshold.

Locked On

This condition occurs if the detector reports 100% occupancy for a duration determined by lane type. It is also sustained if the occupancy drops to zero with no intervening values. The condition will be cleared after 24 hours of good occupancy data.

No Change

If occupancy is greater than zero and does not change for 24 hours, this condition will be triggered. It will clear immediately if the occupancy changes.

Occ Spike

A spike timer is kept for each detector. For every 25% change in occupancy between two consecutive data samples, 30 seconds are added to the timer. If its value ever exceeds 60 seconds, the condidtion is triggered. After every poll, 30 seconds are removed from the timer. The condition will be cleared after 24 hours of no spikes.

Force Fail

If a detector has a fault which is not handled automatically, it can be force failed. This flag is only set manually, so it must be cleared once the failure is corrected.

Fake Detectors

When a detector is failed (auto fail or force fail), its data will not be used for travel time, ramp metering, etc. In that case, fake detection can be used — this field can contain one or more other detector names, separated by spaces. The average density or speed of those detectors (which are not also failed) will be used instead.

Traffic Data Archiving

Sample data is archived only if the sample_archive_enable system attribute is true. Files are stored in /var/lib/iris/traffic, in a directory with the district name. Within that directory a new subdirectory is created for each year, with a 4-digit name (1994-9999).

As data is collected, a new subdirectory is created every day — the name is 8-digits: year 1994-9999, month 01-12 and day-of-month 01-31. At 10 PM, all traffic data from the previous day is moved into a single ZIP file with the 8-digit base name and a .traffic extension.

Vehicle Logging

The .vlog format is a comma-separated text log. Each vehicle event is recorded as a single line of values, ending with a newline \n (U+000A).

Duration is the time a vehicle occupied the detector area, between 1 and 60000 ms. An invalid or missing value is represented by a ? (U+003F).

Headway is the difference in arrival time from the previous vehicle to the current one. It is a positive integer between 1 and 3600000 ms (1 hour). An invalid or missing value is represented by a ? (U+003F).

Time is when the vehicle left the detection area. It is included only when the headway is invalid or missing, otherwise it is left empty.

Speed is the measured vehicle speed. It is a positive integer value from 5 to 120 mph. An invalid or missing value is left empty.

Length is the measured vehicle length. It is a positive integer value from 1 to 255 ft. An invalid or missing value is left empty.

All trailing commas at the end of a line are removed. This means that an event with only duration and headway would only contain the two values, separated by one comma.

A gap in sampling data due to communication errors is represented by * (U+002A) on a line by itself.

Example Log

Interpreting example .vlog data for 11 vehicles:

Binned Data

IRIS can collect these types of binned traffic data:

A binned sample file consists of some number of periods of equal duration. The first period begins (and the last period ends) at midnight. The binning interval determines the number of samples collected per day — a shorter interval results in more samples. If the period is longer than 30 seconds, the samples are allocated evenly into 30-second bins for storage.

For each detector, a binned sample file is created for each traffic data sample type. The base file name is the detector name. The extension is the traffic data code followed by the period (in seconds). For example, 60-second vehicle count samples collected from detector 100 would be stored in a file called 100.v60, containing 2880 bins.

Each data sample is either an 8- or 16-bit signed integer, depending on the sample type. 16-bit samples are in high-byte first order. A negative value (-1) indicates a missing sample. Any data outside the valid ranges should be considered bad.

Traffic Layer

The IRIS client user interface includes a traffic map layer which is created automatically from the road topology. By default, this layer uses traffic density to determine the color of each segment. Other themes are available for speed and flow. The Legend menu at the top of the map can be used to view the thresholds used for each color in a theme.

Every 30 seconds, the client will make an HTTP request for the current traffic data. The URL to locate that file is declared as a property in the /etc/iris/iris-client.properties file (on the IRIS server). The property is tdxml.detector.url, and it should point to the det_sample.xml.gz XML file, as made available by nginx on the IRIS server.

The appearance of the traffic map layer changes depending on the current zoom level. If the zoom level is below 10, the layer will not be visible. At zoom levels 10 through 13, the layer will display segments as aggregate of all detectors in each mainline station. At zoom level 14 or above, each mainline detector will be displayed as a separate segment.

The maximum distance between adjacent stations to draw segments on the map is specified by the map_segment_max_meters system attribute. It is also the maximum downstream distance for associating station data with a segment.