Programs & Labs
Spring 2009
Beyond the loop: Researchers develop the next generation of vehicle detectors
A prototype piezoelectric beam sensor developed by Rajesh Rajamani (right), Krishna Vijayaraghavan, and Lee Alexander is one of the new sensor technologies being studied by the ITS Institute.
In order to manage traffic, you have to know where it is. For nearly 50 years, one of the most widely used tools for measuring traffic flows has been the loop detector— a large coil of electrically charged wire embedded beneath the pavement surface. When a large metal object passes overhead, the coil’s electrical inductance changes, causing a change in the flow of electricity running through it and triggering the detector.
Today, new technologies are available that offer sensing capabilities beyond those of traditional loop detectors—as well as substantially lower costs. With support from the ITS Institute, several independent research groups at the University of Minnesota are developing new sensor technologies that will make tomorrow’s roads smarter.
Introduced in the 1960s, pavement-embedded loop detectors gave engineers a powerful new tool to measure traffic flows. At that time, urban freeway systems were becoming increasingly important to metropolitan traffic, carrying high volumes of commuter traffic between suburban communities and central cities. Over the years, metropolitan areas have installed thousands of loop detectors. More than 6,000 loop detectors are now in place around the Minneapolis-St. Paul region.
The design of loop detectors, however, creates several disadvantages. One significant problem is cost—a single loop detector unit may cost several hundred dollars. Because they are embedded directly in the pavement, loop detectors are difficult to maintain. They require an external power source to operate, so electrical service must be installed along the roadway. When this service fails, detectors are difficult to repair or replace.
Batteryless wireless traffic sensors
A new type of traffic sensor that overcomes many of the limitations of loop detectors is taking shape in the mechanical engineering department laboratory of Rajesh Rajamani. Working with research fellow Lee Alexander and graduate student Krishna Vijayaraghavan, Rajamani is developing a self-powered wireless sensor for highway and arterial use that is more flexible and potentially less expensive than inductive loop systems.
Taek Kwon’s portable wireless sensor nodes could make manual intersection traffic counters like this obsolete.
The sensor consists of a beam embedded in the road surface and a data processing unit installed within a few hundred feet. Piezoelectric materials in the beam convert mechanical energy (in the form of pressure and vibration from cars passing over the beam) into electrical energy, providing enough voltage to power an onboard electronic system that wirelessly transmits a signal to the wireless processing unit. The shape of the piezoelectric beam sensor—a thin bar rather than a large loop—makes it easier to install than a loop detector. There is also no need to connect the beam sensor to power supply lines or data transmission cables.
The narrow shape also unlocks sensing options beyond the capabilities of loop detectors. A piezoelectric unit can detect not only the presence of a vehicle, but can also count each axle that passes over it and determine the length of the vehicle. This information will give traffic managers important new data on what kind of vehicles are using roads.
Wireless mesh sensing
On the University’s Duluth campus, the advantages of wireless sensing are also key to research by Taek Mu Kwon of the electrical and computer engineering department. Kwon directs the Transportation Data Research Laboratory, managing data from the Twin Cities’ freeway loop detector system. But in a recent research project, he turned his attention to the challenge of monitoring complex traffic patterns in arterial intersections.
Engineers analyzing intersection traffic are often forced to rely on manual data collection—workers record the movements of vehicles through the intersection using handheld data loggers, an approach that is both tedious and error-prone. Kwon’s research aims to automate the process with small wireless sensor nodes that are easy to install temporarily on the road surface.
Kwon’s sensor nodes are designed to be installed in groups, with each sensor responsible for detecting vehicles in a single lane of traffic. Once in place, the sensor nodes automatically configure themselves as a “mesh” network, moving the raw data to a processing unit that extracts vehicle trajectories. The sensors are particularly suitable for short-term installations because each one is powered by its own battery and mounted on the road surface with an adhesive, but they can also be connected to an external power source for longer-term applications.
A mesh network is defined by multiple links between nodes. In a mesh topology, data can hop from node to node to reach a destination, rather than being transmitted directly to and from a central point. In a full mesh network, the number of links increases rapidly as more nodes are added. Kwon’s sensor network strikes a balance between flexibility and complexity via a partial-mesh topology, in which each sensor is connected to at least two other nodes in order to provide alternative data routes, but not to every other node in the system.
In operation, each sensor node in an intersection registers the magnetic disturbance caused by a vehicle passing directly over it and transmits the exact time of that event through the mesh network to a data logger positioned nearby. The nodes’ communication protocol ensures that their internal clocks are synchronized, so the timing of every vehicle detection event is recorded accurately.
To turn raw sensor output into usable information, the collected data are processed using a tracking algorithm that reconstructs vehicle trajectories from vehicle detection events. Because individual sensor nodes are responsible for each lane of traffic, individual nodes can be designated as “entrance” and “exit” nodes. Taking into account the geometry of the intersection, the tracking algorithm matches vehicle detection events recorded by entrance nodes with events recorded by exit nodes. The result is a set of node-to-node trajectories representing the movements of individual vehicles.
Carbon nanotube pavements
A very different approach to sensor design is being taken by Xun Yu in the mechanical and industrial engineering department of the University of Minnesota Duluth. Yu’s current research seeks to turn the pavement itself into a sensor by exploiting the electromechanical properties of carbon nanotubes.
Since their discovery nearly 20 years ago, carbon nanotubes—cylindrical carbon molecules in which atoms are organized into hollow cylinders that are only a few atoms in diameter but up to millions of atoms long—have attracted the interest of researchers in many fields due to their unusual properties. In addition to being extremely strong, carbon nanotubes are electrical semiconductors that exhibit linear changes in electrical resistance in response to mechanical stress, a quality known as piezoresistance.
Yu is attempting to put piezoresistance to work by mixing carbon nanotubes (CNTs) with cement. In a wellformulated CNT/cement composite, evenly distributed nanotubes would form a web of carbon filaments spanning the entire paved area. Installing a simple set of electrodes to measure electrical resistance would turn the pavement into a single large pressure sensor.
Though such a sensor design is mechanically simple, with no complex circuits or moving parts, fabricating CNT/cement composites that can perform effectively as sensors is far more challenging than pouring a cupful of nanotubes into a cement mixer. CNTs have an unfortunate tendency to clump together when placed in solution, forming discrete blobs rather than the even, continuous network that piezoresistive sensing requires. To overcome this tendency, Yu is studying different chemical methods of encouraging CNTs to disperse, with an eye toward identifying methods that can be incorporated into commercial concrete mixing processes.
In addition to developing a manufacturing process for CNT/cement composites, Yu’s sensor concept also depends on developing a thorough understanding of their electrical and mechanical properties. In the laboratory, Yu is currently investigating composites’ piezoresistive response to dynamic and static stresses, as well as the effects of temperature, humidity, and other environmental factors.