DRI Laser Detector
Detailed Information
How it works
This detector is quite a bit different from other laser based traffic detectors. Conventional Lidar based laser detectors use the time of flight of light to figure out how far objects are away. Light moves rather fast, so this timing must be quite precise, which is difficult under the wide range out outside temperatures and conditions. Scanning Lidar detectors scan across the roadway by rotating a beveled mirror. The timing on the mirror movement must be ultra precise, which is difficult on a typical mount which is not perfectly stable. The DRI Laser Detector mitigates these limitations of conventional laser based traffic detectors.
The basic concept of the DRI Laser Detector is that invisible infrared laser light shines down on the roadway and a modified optical telescope which is offset from the laser looks for that reflection off the pavement. Due to the physical offset of the laser and return optics in the detector, and the high precision of the laser, any vehicle greater than a foot or so high will block this reflected beam path, and hence be "detected." The diagram below was used in our original patent application and shows the concept of "critical height," where any object greater than this critical height will be detected because it disrupts the light path reflected between the laser and receiver.
Basic Laser Detector Concept
In actual practice, the detector has a set of two lasers and optical receivers - one in the front, and one in the back. These two lines of detection a known distance apart are used to determine speed. The picture below shows the physical layout of the detector.
Laser Detector Layout
The Basic System Operation shows how the basic principal of the detector is applied in operation. The two sets of laser lines and return optics a known distance apart are used to determine the speed, and from the speed and residence time are determined the precise vehicle length. Because the laser is sampled a 10 kHz, we can determine the vehicle's length to within about a cm when it is traveling at 70 mph, and more accurately at slower speeds. This ultra-precise vehicle outline is indicative of the vehicle class.
This file shows the operation of the detector from a top down view. A laser line rather than just a single point is used to capture the vehicle independent of its location in the lane. Within the return optics there are a number of discreet sensors, providing redundancy in the speed detection, and therefore further accuracy. The bottom of the screen shows an experimental user interface which outputs the speed and length determined by each pair of detectors. It can be seen that the detector elements at the center of the vehicle fire first because the vehicle is longer at its center due to the bumper curvature.
Because Caltrans currently has no Standard Plans allowing the placement of electronic devices over live traffic, DRI and UC Davis have also developed a new mounting system which will make the deployment of this detector much quicker, simpler. safer, and less expensive. This new Winch-and-Trolley Based Mounting System allows the exact placement of this (or any other) detector on overhead structures above live traffic. This mount allows the detector to be raised and lowered from the ground by a single field worker in a few minutes without requiring a bucket lift truck. This mounting system is an adjunct to the existing Standard Plans, and was designed to be available as a Standard detail add-on to existing Caltrans overhead structures by the simple addition of two aluminum C channels.
The operation of this new mounting system is quite easy, as shown in these videos: [High Resolution] / [Low Resolution]. The detector is attached to a trolley which is raised by a winch and automatically aligns to two C channels. The trolley then goes out over live traffic, cinches down, and the detector is powered from the channels. An operational demonstration of a prototype of this mounting system, with a live laser detector, was given to Caltrans on a UCD Test Site, as seen here. UCD will have another demonstration at this safe, no-traffic site during the summer of 2005.
The safety of this device was assured long before the first prototype. The power output is much, much less than sunlight. The laser detector was evaluated by University of California laser safety experts and deemed to have no effect on the public or researchers using it. This detector outputs around 4 mW, which is about the power output of an indoor laser pointer, only this laser output is spread straight out over a 12-foot area. This is much less power than, for example, law enforcement shines directly at drivers for Lidar based speed enforcement.
This detector provides highly accurate reliable traffic data without embedding anything in the roadway. It makes use of a mounting system that can precisely place the detector on existing overhead structures without requiring a lift bucket truck. It not only generates speed, volume, and occupancy data, but provides the precision and reliability to re-identify vehicles between sites, providing travel time and origin/destination information as well.
The detector uses a multiplicity of ultra high resolution sensing elements, allowing the determination of the vehicle bumper curvature. The vehicle bumper curvature ends up being another very useful characteristic to differentiate individual passenger vehicles. High density commuter vehicles are approximately the same size, but the combination of their exact size and bumper curvature makes them quasi-unique. Re-IDing passenger vehicles has been virtually impossible during commute periods when most vehicles have very similar characteristics, necessitating this ultrahigh precision detector.
The detector uses lasers for detection. The California Air Resources Board is also using lasers to determine the pollutant emissions of vehicles at single lane onramps. There are plans to modify the configuration and wavelength of this laser detector so that it too can measure pollution, although this detector would measure it from the top down for all lanes. This would be a major advantage over the currently pollutant sensors which require elements on both sides of the roadway and therefore can only measure one lane at a time.
The best way to understand this detector is to see it in operation.
Later this summer, DRI is planning a roadway demo of the laser detector and mounting system on an isolated part of the UCD campus. This site is ideal because parking is easy and the operation of both laser and mounting system can be seen up close and personal in a safe environment. Although the exact date hasn't been set yet, if you are interested in attending, contact Joe Palen at japalen@dcn.org.
By early next year, DRI will have built both the laser detector and mounting system on a sign truss to be installed this summer in Sacramento. Hence, it will be possible to see the the detector operational over live traffic. Although there will be a small protected parking area, this site is not large enough and is too noisy for a formal demonstration. Hence, anyone interested in this detector and/or mounting system should check out the UC Davis demo, which will be a better and safer place to see it.
Another way to understand this detector is to view its operation using VideoSync, a traffic detector analysis tool specifically developed to link detector data with the associated traffic video. The VideoSync player is available at this link. Additionally, you will need the laser data in the directory referenced below. VideoSync syncs the data from a detector with concurrent video, and it is used primarily to validate and "fix" conventional loop and radar detectors, but it can be used to assess any type of detector. VideoSync allows playing of the video frame by frame or at any speed and can automatically jump to vehicle detection events, freezing the video at the point of detection.
The data directory in the test1.zip contains a short QuickTime video clip of freeway traffic, with which VideoSync syncs the laser data, visually demonstrating operation. This video has not been compressed, so it may take a long time to download over low bandwidth links. This video clip shows the unit detecting a number of vehicles, some where it can be seen that the protruding parts of the bumpers are detected before the rest of the vehicle, exemplifying the speed of the detector. In all cases, the detector does detect the curvature or shape of the bumper as a distinguishing vehicle characteristic; unfortunately, the video is only captured at 30 frames a second, which is about 0.3% the speed of the detector, so this phenomenon is not visually apparent in all cases. Also at 70 mph (+), vehicles are somewhat blurry in the video, but it is still possible to catch the general gist of its operation. It also should be noted that VideoSync reduced the laser detector data from 10 kHz to the standard Log170 rate of 60 Hz, so there are only two data time steps for every one video frame in VideoSync, even though there are more than 300 in the raw laser data.
For those with software or bandwidth restrictions that don't allow the ready loading of software, here's what you'd see. The two rows of translucent boxes are the detector elements, numbered from 0 to 7 (the production version of the detector will have two rows or 24 detection elements extending across the full lane). As the vehicle approaches, all the numerals in the detection elements are colored black, indicating no detection.
VideoSync Vehicle Detection
Time Step #1
At the next Videosync time-step (1/60th of a sec later), upstream detection elements 4 through 7 have turned white because the curved bumper has intersected them first.
VideoSync Vehicle Detection
Time Step #2
At the next VideoSync time-step, the video frame advances (the car has moved) and now upstream detection elements 1 through 7 have fired, as well as downstream elements 4 through 7.
VideoSync Vehicle Detection
Time Step #3
In the next two VideoSync time steps, we can see that the all elements have fired because the vehicle is intersecting all of the detection elements.
VideoSync Vehicle Detection
Time Step #4
VideoSync Vehicle Detection
Time Step #5
Note: This is much easier to see and understand using the VideoSync player, so if you have the bandwidth to download the sample files, that is highly recommended.
Up to this date, the detector has not undergone 24/7 freeway testing because there has been no site where it could readily be mounted in a secure location. Only one prototype was initially built, and we didn't want to loose it to theft or vandalism.
Concurrent with the construction of the new secure test site in Sacramento the summer of 2005, and the development of the associated raising/mounting mechanism, we will be building a number of these detectors in an environmentally hardened form factor for continuous outdoor testing through the winter of 05/06. This will allow Caltrans to assess the long term stability of the design.
Pending this evaluation, there are plans to:
- reduce the power consumption so the detector can operate solely on a combination of solar and fuel cells;
- add jerk (nonlinear acceleration) detection to the unit, this being necessary to get accurate travel time in stop-and-go traffic conditions; and
- add pollution and other hazardous chemical species detection by changing the laser wavelength used.