Incidents Showing Up on Radar

The basic problem with incident detection has been, and continues to be, detectors. If you look at a freeway map of major metro areas in California (such as displayed at www. dot.ca.gov/traffic/), you will see that most sites along freeways show a status of “No Data Available.”

These sites are not reporting traffic data as a consequence of detector failures, and the detectors are wire loops in the road paving. This makes California drivers, like myself, rather untrusting of freeway information sites.

While it might be desirable to obtain travel time data from freeways by using inductive signatures of vehicles, this approach is dependent upon loops of wire embedded in the roadway, just like the wire loops we already have in place that aren’t working. Loops require lane closures for installation, and can create congestion in an area like Los Angeles even when installed at 2 AM. Even if lane closures are tolerable, loops fail when pavement fails, making the installation of loops a recurring maintenance task.

Maintenance issues and lane closure issues aside, there are accuracy issues with loops. Loop detectors are not able to distinguish between two vehicles in adjacent lanes versus one vehicle changing lanes. And, when loop detectors are adjusted to have enough sensitivity to detect motorcycles, buses are often detected in adjacent lanes, as well as in the lane in which they are traveling. The maintenance and accuracy weaknesses of loop detectors should motivate the traffic industry to seek other detectors.

Moving above ground

Over the years, many above-ground vehicle detectors have been developed and brought to market as products. Video detection, radar and laser scanners are among these products.  Video detection and laser scanners require equipment installed over the roadway, which poses maintenance and lane closure issues nearly as serious as loop detectors. Radar is the only above-ground detector that has been able to address all of the safety, lane closure, maintenance and accuracy issues.

Radar is installed at the side of the right-of-way and can monitor eight separate lanes with nothing in or on the roadway. Radar also is a 50-year-old technology that can be implemented using solid-state electronics. When implemented using solid-state electronics, radar easily achieves a 10- to 20-year mean time between failures and eliminates the need for periodic maintenance. Notwithstanding all of its pluses, radar has not been widely embraced by freeway surveillance systems because it is viewed as “new technology.”

For those of us old enough to remember, it took a long time for loops to be accepted by the traffic industry. Radar will not be much different.

If freeway detectors had the reliability and accuracy of radar, the freeway maps for major metro areas would show the actual status of traffic, rather than the “No Data Available” status which is the rule for most sites today. With reliable and accurate freeway traffic volume, speed and percent lane occupancy—like that which is available from radar—traffic management and incident detection using volume, speed and lane occupancy data would be practical and straightforward. To manage traffic, one must know where the traffic is and how fast it is moving. Unfortunately, most transportation agencies cannot obtain real-time traffic data because such a large number of their loop detectors aren’t working. They are managing traffic without real-time data for most of their network of roadways. It is no wonder that ramp meters run on the basis of time of day and day of week, rather than on the basis of admitting traffic at a rate that doesn’t overload the freeway.

Time as detection?

There is a contention that travel time can be used as an incident detection tool. This overlooks a significant issue. One must wait until vehicles travel from one detection station to another before travel time can be measured. If there is an incident between the two stations, the wait for travel time data will be unreasonably long and will delay response to an incident. Considering this delay is acceptable, travel time measurements that rely upon the inductive signatures of vehicles puts users back in the mode of being dependent upon the same loop detectors that are not working properly at most sites today. Travel time measurements using automated license plate reading are much more reliable, both in terms of identifying the vehicle and in terms of using equipment that is out of the pavement and durable for decades. License plate reading does have the drawback of arousing concerns about privacy.

It is possible that incidents can be detected more efficiently and less intrusively without travel time. This can be accomplished by using the first derivative with respect to time of common traffic measurements, such as volume, speed and occupancy.

When speed declines substantially and occupancy increases substantially at the same site during a brief time period, such as one minute, these changes can signal an incident. This should cause a CCTV camera to be oriented toward the roadway segment to determine the cause for this effect. If there is an incident, visual verification via CCTV can be done before the growing queues of vehicles extend to an upstream traffic monitoring site.

For those wanting more thorough automatic incident detection, traffic flow volume as measured upstream is not likely to be maintained at sites downstream of the incident (incidents are accompanied by queues of slowing or stopped vehicles, aren’t they?). With vehicle detectors counting traffic movements on off-ramps and on-ramps, changes in freeway volume could be used to further confirm an incident and activate a CCTV view of the roadway just upstream of the site where flow volume has declined precipitately. However, using this additional effect will also take time, and this time will add to delays in detecting the incident and dispatching resources. The Freeway Service Patrol may arrive at the incident site before this further degree of automation causes the incident to be detected.

Out of the loop?

A basic solution may work better: provide adequate traffic data for the entire network of freeways. Automatic incident detection is entirely dependent upon adequate traffic data. Adequate traffic data means measurements of vehicle flow volume per lane, average vehicle speeds per lane and lane occupancy at every 0.25 mile or 0.5 mile along the entire length of the roadway network. It also means data that doesn’t require corrections for errors from sites where detectors in some lanes aren’t working. Adequate traffic data is, and will always be, a critical enabler for automated incident detection.

What prevails today is inadequate data from just a part of the roadway network. This makes automated incident detection difficult, if not impossible. The result is complicated software in traffic management centers that attempts to detect incidents by reviewing incomplete data, and this data also contains errors attributable to loop detectors that, while still working, are not working well and may be near to failure.

The first step in solving a problem is to admit to having a problem. Too many freeway traffic maps show a large percentage of sites with “No Data Available.” It is time for the traffic industry to admit that loop detectors are the problem and that loops should be abandoned in favor of above-ground detectors that work well from readily accessible locations, like the edge of the right-of-way.

Radar is one of these detectors. There are probably other technologies that can be used, as well.  But the traffic industry must first admit to its problem. Then, and only then, will reliable freeway traffic sensors be made and generally deployed that can deliver the adequate traffic data needed to make automatic incident detection feasible.

Until then, drivers like myself will continue to tune in KNX1070 or KCBS to learn where the problem spots are in Los Angeles or San Francisco. People with cell phones do a pretty good job of calling these radio stations promptly to report incidents and have done so for nearly 15 years. These radio stations are the state of the art in incident detection today, and advise drivers within minutes of the occurrence of an incident. The freeway traffic maps from websites like www.dot.ca.gov/traffic do not compete with the quality and scope of information provided by these AM radio stations.   TME