The smart road is an experimental highway currently under construction in Virginia. It is being built from the ground up with ITS testing and evaluation in mind. Embedded research support infrastructure will include underground conduits, underground junction bunkers, power, a fiber optic data network, embedded pavement sensors, snow making capability, and experimental lighting. The facility will be utilized for a number of research areas including safety and human factors, snow and ice control, pavement research, bridge and structures research, ITS sensor development and evaluation, and roadside to vehicle communications.

THe primary objective for this project is the design of optical fiber-based sensor instrumentation for specific ITS applications. Specifically, this paper discusses research on optical fiber sensors that can be used for traffic monitoring and vehicle classification. This paper also discusses developments on the application of optical fiber sensor that can be used for monitoring visibility. This research is directly beneficial to the implementation of driver advisory and safety systems, traffic control system, and other ITS applications. This paper summarizes research performed on optical fiber sensors used for measuring traffic flow on highways and discusses progress on optical fiber sensors used for monitoring visibility.

Electronic Toll Collection (ETC) via Automatic Vehicle Identification (AVI) technology has significantly altered traffic operations during toll collection. In particular, the value of the average processing rate of a lane providing both ETC service as well as a traditional service, fluctuates over the rush hour between the average value of the processing rate of the traditional service and the capacity of the ETC service. This study develops a queuing model to address the changing processing rates for the different mixed lanes. The model is applied to the westbound 9-lane portion of the Holland East Plaza in Orlando, FLorida. Data is evaluated for 6 different rush hours that include 3 different configuration patterns implemented over a period of 3 years. In the first configuration, only the traditional toll collection services are provided. In another configuration, all traditional lanes become mixed to include ETC except for the center lane, which becomes a lane dedicated solely to ETC service. In a final configuration, two lanes become dedicated to ETC service.

Calspan SRL Corporation is currently developing an on- vehicle threat detection system for intersection collision avoidance (ICA) as part of its ICA program with the National Highway Transportation Safety Administration. Crash scenarios were previously defined and an on-board radar sensor was designed. This paper describes recent efforts that include the development of a simulation of a multitarget tracker and collision avoidance algorithm used to predict system performance in a variety of target configurations in the various ICA crash scenarios. In addition, a current headway radar was mounted on the Calspan Instrumented Vehicle and in-traffic data were recorded for two limited crash scenarios. Warning functions were developed through the simulation and applied to the recorded data.

Motion detection, for collision avoidance,using VLSI monolithic smart sensors that mimic insect vision have ben reported for some time. Due to the parallel processing required by insect vision, actual chip implementations have tended to be limited to 1D arrays of integrated photodetectors, or modest 2D arrays. This paper reviews our progress, to date, and examines some of the issues that lie ahead for large 2D insect vision arrays.

Laser scanners, or ladars, have been used for a number of years for mobile robot navigation. Although previous scanners were sufficient for low-speed navigation, they often did not have the range or angular resolution necessary for mapping at the long distances required by high-speed navigation. Many also did not provide an ample field of view. In this paper we will present the development of state-of-the-art, high speed, high accuracy, laser range scanner technology. This work has been a joint effort between CMU and K²T in Pittsburgh and Zoller + Friehlich in Wangen, Germany. The scanner mechanism provides an unobstructed 360 degrees horizontal field of view, and a 30 degree vertical field of view. Resolution of the scanner is variable with a maximum resolution of approximately 0.06 degrees per pixel in both azimuth and elevation. The laser is amplitude-modulated, continuous-wave with an ambiguity interval of 52 metes, a range resolution of 1.6 mm, and a maximum pixel rate of 500 kHz. This paper will focus on the design and performance of the scanner mechanism and will discuss several potential applications for the technology. One application, obstacle detection for automated highway applications will be discussed in more detail. Example data will be shown and current mechanism improvements from the CMU prototype will also be discussed.

In order to establish mobile robot operations and to realize survey and inspection tasks, robust and precise measurements of the geometry of the 3D environment is the basis sensor technology. For visual inspection, surface classification, and documentation purposes, however, additional information concerning reflectance of measured objects is necessary. High-speed acquisition of both geometric and visual information is achieved by means of an active laser radar, supporting consistent range and reflectance images. The laser radar developed at Zoller + Froehlich (ZF) is an optical-wavelength system measuring the range between sensor and target surface as well as the reflectance of the target surface, which corresponds to the magnitude of the back scattered laser energy. In contrast to other range sensing devices, the ZF system is designed for high-speed and high- performance operation in real indoor and outdoor environments, emitting a minimum of near-IR laser energy. It integrates a single-point laser measurement system and a mechanical deflection system for 3D environmental measurements. This paper reports details of the laser radar which is designed to cover requirements with medium range applications. It outlines the performance requirements and introduces the two-frequency phase-shift measurement principle. The hardware design of the single-point laser measurement system, including the main modulates, such as the laser head, the high frequency unit and the signal processing unit are discussed in detail. The paper focuses on performance data of the laser radar, including noise, drift over time, precision, and accuracy with measurements. It discusses the influences of ambient light, surface material of the target, and ambient temperature for range accuracy and range precision. Furthermore, experimental results from inspection of tunnels, buildings, monuments and industrial environments are presented. The paper concludes by summarizing results and gives a short outlook to future work.

THe purpose of this paper is to show to what kind of application laser range-finders can be used inside Autonomous Intelligent Cruise Control systems. Even if laser systems present good performances the safety and technical considerations are very restrictive. As the system is used in the outside, the emitted average output power must respect the rather low level of 1A class. Obstacle detection or collision avoidance require a 200 meters range. Moreover bad weather conditions, like rain or fog, ar disastrous. We have conducted measurements on laser rangefinder using different targets and at different distances. We can infer that except for cooperative targets low power laser rangefinder are not powerful enough for long distance measurement. Radars, like 77 GHz systems, are better adapted to such cases. But in case of short distances measurement, range around 10 meters, with a minimum distance around twenty centimeters, laser rangefinders are really useful with good resolution and rather low cost. Applications can have the following of white lines on the road, the target being easily cooperative, detection of vehicles in the vicinity, that means car convoy traffic control or parking assistance, the target surface being indifferent at short distances.

Two solutions to the vehicle location problem are commonly discussed for Intelligent Transportation Systems (ITS): active roadside beacons and global positioning system (GPS) satellites. This paper present requirements for new linear referencing sensors, defined as sensors that will identify a vehicle's location along a roadway in terms of distance along the roadway from known points or by the automatic identification of known points. Requirements for linear referencing sensors come from new national location referencing standards being developed by initiatives of the US Department of Transportation, and from international location referencing standardization activities. Linear referencing sensors can extract information from the visual scene presented by the roadside environment, or from the environment illuminated by laser or microwave radiation. They can also be based on new, low cost techniques for labeling roads or by modulating lane reflectors or other regular road infrastructure components. Such sensors, singly and in combination, avoid the map matching problem common to vehicle navigation systems that rely on GPS, and can be deployed at much lower cost than roadside beacons, particularly when designed as one function of multi-purpose in-vehicle sensors and computers.

We have built a portable spectro-polarimetric machine vision system that operates at video frame rates. Our system contains only electronically controllable components, including an imaging acousto-optic tunable filter (AOTF), a phase retarder, acceptance and imaging optics, and a standard CCD-based camera. The device operates like an ordinary camera, except that a computer controls the spectral and polarization content of light to be viewed. For example, by sweeping the wavelength over the AOTF's range, one can obtain a spectral signature for each pixel in an image. Alternately, the camera can switch between two wavelengths, allowing for high-speed discrimination of closely matched colors in a scene. In addition to digitally controlling the wavelength, our imager uses a liquid crystal retarder to filter images based on polarization signatures of objects. We have implemented a number of algorithms to take advantage of the unique capabilities of our sensor, some of which can be applied to problems specific to transportation systems. We present two image processing applications that highlight the different methods we use to analyze scenes with our system. One application uses spectral processing to locate vegetation in a scene; the second uses polarization signatures to detect glare from hazardous road conditions such as water and ice.

A novel motion detector utilizing a millimeter-wave array front-end, with signal processing that mimics insect vision, is described. The use of passive millimeter-wave detection enables a significant improvement over optical or IR wavelengths, when a colliding object is obscured by rain, steam or other aerosols. This, for instance, used as a blind-spot detector, will enhance driver safety in poor weather conditions. As insect vision techniques do not attempt to process an image, but rely on tracking moving edges, the processing tasks are less hardware intensive, resulting in a compact low-cost solution.

Microscopic traffic simulators capture the effects of driver behavior and so provide results not available from other types of traffic simulation. In order to fulfil this potential though, the behavior related models in the simulator must use parameters based on real measured driver behavior. We have turned to analyzing locally collected video traffic data to support our simulation of a new network in Boston. We used a recently developed software tool to extract the driver behavior information. This paper discusses the important tradeoffs in camera sitting and how they affect the subsequent data processing.We also describe the calibration procedures and field measurements necessary to extract information using this software. We present some of the driver behavior data we developed in this effort. The procedures described here can successfully generate the driver behavior information to support accurate microscopic modeling of vehicular traffic.

This paper considers an active vision system with spaced light source and receiver that simulates illumination and observation conditions typical for a car driver in fog. The objects of the investigations are image contrast and limiting visibility range for a characteristics non- Lambertian target, namely for a retroreflector. Calculations are made for a Gaussian angular dependence of its radiance factor centered near the direction opposite to the incident light direction. We concentrate our attention on studying effects of the angular retroreflectance scale on contrast and limiting visibility range of such a retroreflective marker. Here is shown the optimal retroreflective characteristics to exist that provide the maximal contrast and the best visibility of the marker under other equal conditions. The estimations of the angular scale are presented for various practical situations. The limiting visibility range values of retroreflective and Lambertian markers are compared.

The North American Free Trade Agreement is anticipated an already increasing trend in highway freight movement across the international border between Texas and Mexico. The Texas Department of Transportation is concerned about safeguarding its motoring public and protection of its highway infrastructure. By sponsoring this research, it hopes to improve traffic signal operations, especially at isolated intersections. Recent advances in sensing technologies and signalization enable safer and more efficient intersection control. This research evaluated advanced detection technologies that can be used to increase green time to trucks and reduce their stops and delays. Equipment selection criteria required devices that were reasonably accurate in classification of vehicles under all weather and lighting conditions and determination of vehicle speeds. The research team selected active IR and passive acoustic technologies. Components of the detection system include: an Industrial PC, proprietary boards inside the computer, IR and acoustic detectors, a pole for mounting the selected systems, and a classifier system using pavement sensors for verification purposes. The purpose of the node computer was to interpret signals from detectors, store data, and communicate with the controller cabinet upon detection of a truck.

To address the serious problem of truck rollover accidents on freeway exit ramps a system was developed and implemented by the Federal Highway Administration (FHWA), a private consultant Bellomo-McGee, and a system integrator International Road Dynamics. The system utilizes several existing technologies to determine vehicle weight, vehicle type, vehicle speed, and vehicle declaration. The system uses the information gathered to evaluate each vehicle on a freeway exit ramp to determine if they are in danger of a rollover accident and provides a warning to vehicles in potential danger. The system was implemented at three sties in the Washington DC area that had a history of rollover accidents. A three year independent evaluation was conducted on behalf of the FHWA to determine the effectiveness of the system. The evaluation shows that the system has been effective in reducing speeds and reducing accidents at the three sites that were chosen and shows that the systems are economically beneficial.

In automatic highway systems, automatic brake actuation is a very important part of the overall control of the vehicle. Hence, a faster response and a robust braking system are crucial. This paper describes electromagnetic brakes as a supplementary system for regular friction brakes. This system provides better response time for emergency situations, and in general keeps the friction brake working longer and safer. A new mathematical model for electromagnetic brakes is proposed to describe their static characteristics. The performance of the new mathematical model is better than the other three models available in the literature.

This paper describes an approach we used for designing an automotive obstacle detection system. This system will be able to detect hazardous situations in road traffic by the data fusion of a set of sensors and to warn the driver when such situations occur. Our approach can be summarized in three points: (1) We used a distributed architecture of smart sensors which are binded by a dedicated data network. (2) Since the experiments on a test tack are often difficult, our system makes possible the recording of all sensor data, through the network, in real time. These data are time stamped and digital video images are also recorded at a frequency of 2 Hz. Then, it becomes possible to make long recording sessions of different road scenarios and to have a pool of data at our disposal for testing and validating obstacle detection algorithms. (3) We use a classification of the sensors to make connection of new sensors to the network easier.

This paper is concerned with the design of a feedback controller to minimize accumulated queue lengths in the presence of unknown inflow disturbances at an isolated highway intersection and a simple network with two intersections. The admissible control set is taken to be a finite set at each point of time. The value function for the associated game is obtained as the solution of the appropriate Hamilton-Jacobi-Isaacs equation.

A nonlinear programming model was formulated to provide an integrated traffic control strategy for a freeway corridor under incident conditions. The model includes diversion routes, diversion rates, on- and off-ramp metering rates, and arterial intersection timing plans as control variables. A gradient projection algorithm was employed to solve simultaneously the optimal control measures. The model performance was evaluated and validated by running the simulation and optimization programs of TRANSYT-7F and INTEGRATION. It has been found that the proposed model and control strategy reduce the overall system delay, increase the throughput of the corridor, and thus improve the traffic conditions of the entire corridor.

Both recurring and non-recurring congestion on the nation's urban freeways have become a major problem threatening the economic growth of many urban centers. One of the potential solutions to this continuously worsening traffic congestion problem is the control of the influx of traffic to the freeway through the use of ramp metering strategies. The concept of ramp metering implies that the amount of traffic entering a freeway from on-ramps will be regulated in such a way that the traffic will not exceed freeway capacity.

SmartAHS is a specification, simulation, and evaluation framework for modeling, control and evaluation of Automated Highway Systems (AHS). SmartAHS is developed using SHIFT, a new programing language with simulation semantics. This paper discusses the requirements that have led to the development of SmartAHS and SHIFT, summarizes the main characteristics of the SHIFT language, describes the components of the SmartAHS framework and its application methodology. An example - implementation of a tactical level human driver model - is presented in detail; it is sued as a foundation in the studies and evaluation of the partial automation concepts.

This paper present learning stochastic automata as a model for traveller route choice. The theoretical aspects of the model are discussed. A JAVA based program was developed to acquire data from subjects, which are being used to validate the models.

Rising with the expansion of today's transportation systems are needs for new techniques to handle the increasing demand load. This includes the development of Intelligent Transportation Systems (ITS) to incorporate modern technology into creating flexible transportation systems which react more efficiently to traffic problems. A particular focus is the creation of an Automated Highway System (AHS) to combine advanced sensing and communication technologies to create highly efficient computer-controlled traffic flow. At present, many complex dimensions of automated highway development remain difficult to mange. To develop an evolutionary step toward an automated highway, increasing traffic flow in a highly controlled environment is essential. The Lincoln Tunnel, a Hudson River crossing between New York City and New Jersey, represents a major artery to thousands of commuters living in New Jersey. The Port Authority of New York and New Jersey, the governing authority of the Lincoln Tunnel, has dedicated certain priority lanes for bus traffic to streamline traffic flow and create a reward process for mass transit commuters. It is possible to increase flow efficiency using existing technology to control vehicle motion through this corridor. This paper provides a description of physical simulation testing the feasibility of automating lanes for bus flow on this roadway.

Sensor technology plays a critical role in the operation of the Automated Highway System (AHS). The proposed concepts depend on a variety of sensors for positioning, lane- tracking, range and vehicle proximity. Since large substations of the AHS will be designed and evaluated in simulation before deployment, it is important that simulators make realistic sensor assumptions. Unfortunately, the current physical sensor models are inadequate for this task since they require detailed world state information that is unavailable in a simulated environment. In this paper, we present an open-ended, functional sensor hierarchy, incorporating geometric models and abstract noise characteristics, which can be used directly with current AHS tools. These models capture the aspects of sensing technology that are important to AHS concept design such as occlusion, and field of view restrictions, while ignoring physical-level details such as electromagnetic sensor reflections. Since the functional sensor models operate at the same level of granularity as the simulation platform, complete integration is assured. The hierarchy classifies sensors into functional groups. The models at a particular level incorporate characteristics that are common to all sensors in its subgroups. For example, range sensors have a parameter corresponding to a maximum effective range, while lane-trackers include information pertaining to lateral accuracy.

The objective of our work has been to develop and integrate prediction, control and optimization modules for use in highway traffic management. This is accomplished through the use of the Semantic Control paradigm, implementing a hybrid prediction/routing/control system, to model both macro-level as well a micro level. This paper addresses the design and operation of a Kalman filter that processes traffic sensor data in order to model and predict highway traffic volume. This data was given in the form of hourly traffic flow, and has been fit using a cubic spline method to allow observations at various time intervals. THe filter is augmented via the Method of Sage and Husa to identify the parameters of the system noise on-line, and to determine the dynamics of the traffic process iteratively to aid in the prediction of the future traffic. The results show a good ability to predict traffic flow at the sensors for several time periods in the future, as well as some noise rejection capabilities.

Dynamic traffic origin-destination estimation has received increasing attention in recent years due to its grate technological and economic advantages. In this paper, a Kalman filter model for estimating and updating dynamic OD flows for beltway network is presented. The uniqueness of the model is the use of path flows as state variables instead of OD flows, this treatment avoids the likely biases inherent in various traffic assignment procedures. Preliminary testing of the model shows encouraging results, however, more extensive testing is still necessary. Simulation output is used in generating time dependent traffic mapping matrix, additional studies need to be conducted to find out if this would be applicable to the scenario of real-world of considerable size.

This paper discusses the problem of delay estimation in the context of incident management problem. It then provides the assessment of the effects of the timings of the dynamic lane clearance during traffic accidents. The data collection efforts conducted in Northern Virginia is also described. Finally, a simple case study is presented in order to demonstrate the difference between suing and not using the sequential lane clearance information while estimating the traffic delays due to the incidents.

Incident response strategies deal with the resource allocation and minimum risk re-routing of hazardous material carriers in response to traffic incidents. Such strategies also address the effect of loss in coverage due to the non- availability of response vehicles. In addition, the effect of an incident on hazmat carriers that are presented in the local vicinity is that the resulting traffic congestion considerably increases the risk of exposure for motorists. In such a situation, the hazmat carrier needs to be re- routed through a safer path. We have proposed mathematical models that address each of the two problems. The incident response model identifies an optimal allocation of response agency units and their routing in the even of traffic incidents in real-time, while also taking into consideration the response requirements for future demands. The hazmat re- routing model determines an optimal diversion route in a network using multiple risk-based objective functions to re- route Hazmat carriers in the event of an incident occurring along the prescribed route. Mathematical programming formulations are presented for each case, and solution procedures are outlined. Some computational experience is also provided.

We present a computer vision system for detection and recognition of traffic signs. Such systems are required to assist drivers and for guidance and control of autonomous vehicles on roads and city streets. For experiments we use sequences of digitized photographs and off-line analysis. The system contains four stages. First, region segmentation based on color pixel classification called SRSM. SRSM limits the search to regions of interest in the scene. Second, we use edge tracing to find parts of outer edges of signs which are circular or straight, corresponding to the geometrical shapes of traffic signs. The third step is geometrical analysis of the outer edge and preliminary recognition of each candidate region, which may be a potential traffic sign. The final step in recognition uses color combinations within each region and model matching. This system maybe used for recognition of other types of objects, provided that the geometrical shape and color content remain reasonably constant. The method is reliable, easy to implement, and fast, This differs form the road signs recognition method in the PROMETEUS. The overall structure of the approach is sketched.

In independent vehicle concepts for the Automated Highway System (AHS), the ability to make competent tactical-level decisions in real-time is crucial. Traditional approaches to tactical reasoning typically involve the implementation of large monolithic systems, such as decision trees or finite state machines. However, as the complexity of the environment grows, the unforeseen interactions between components can make modifications to such systems very challenging. For example, changing an overtaking behavior may require several, non-local changes to car-following, lane changing and gap acceptance rules. This paper presents a distributed solution to the problem. PolySAPIENT consists of a collection of autonomous modules, each specializing in a particular aspect of the driving task - classified by traffic entities rather than tactical behavior. Thus, the influence of the vehicle ahead on the available actions is managed by one reasoning object, while the implications of an approaching exit are managed by another. The independent recommendations form these reasoning objects are expressed in the form of votes and vetos over a 'tactical action space', and are resolved by a voting arbiter. This local independence enables PolySAPIENT reasoning objects to be developed independently, using a heterogenous implementation. PolySAPIENT vehicles are implemented in the SHIVA tactical highway simulator, whose vehicles are based on the Carnegie Mellon Navlab robots.

The Autonomous Vehicle Team of Virginia Tech is an undergraduate and graduate research and design project in the College of Engineering at Virginia Tech. The goal of the team is self-education and research in the area of autonomous vehicles and navigation system. The team also hopes to achieve the complementary goal of wining the annual unmanned ground robotics competition, which is sponsored by the Association of Unmanned Vehicle Systems International. The competition rules require that the entrants compete in two events. The design competition is an assessment of the vehicle control system architecture, component design and implementation, and overall vehicle design. The dynamic competition requires that the autonomous vehicles circumnavigate an outdoor obstacle course. Vehicle performance is evaluated based on the distance travelled through the course and the ability to avoid obstacles. In order to build effective autonomous vehicles and enrich the interdisciplinary curriculum of the university, the Autonomous Vehicle Team of Virginia Tech is composed of faculty advisors and undergraduate and communications studies. Team members may choose to join more than one of four major technical subgroups of the team and thus gain exposure to technologies they would otherwise not encounter in the conventional curriculum of their respective departments. This paper is an account of the education and research of the interdisciplinary team and the means by which these activities are executed.

This paper presents the setup for conducting experiments, and some results obtained in these experiments, on scaling issues between scaled models and full scale vehicles as they are related to road-tire interactions. The work presented in this paper has profound implications on the applicability of the flexible low-cost automated scaled highway laboratory being developed at Virginia Tech. The laboratory is a proposed 1/15th scale hardware working model of Automated Highway Systems. The vehicles are equipped with ultrasonic sensors for longitudinal guidance. IR and vision system for lateral guidance. They are controlled by HC11 microprocessor boards, and have wireless two way communication infrastructure for vehicle-highway communication. This paper describes the hardware, software, and design issues for the experimental setup and results of some preliminary experiments conducted on scaling issues which provide encouraging results.

We present a real-time model-based vision approach for detecting and tracking vehicles from a moving platform. It was developed in the context of the CMU Navlab project and is intended to provide the Navlabs with situational awareness in mixed traffic. Tracking is done by combining a simple image processing techniques with a 3D extended Kalman filter and a measurement equation that projects from the 3D model to image space. No ground plane assumption is made. The resulting system runs at frame rate or higher, and produces excellent estimates of road curvature, distance to and relative speed of a tracked vehicle. We have complemented the tracker with a novel machine learning based algorithm for car detection, the CANSS algorithm, which serves to initialize tracking.