1

MMF 320 Active Saftey 2005

International Masters Programme in Automotive Engineering Department of Machine and Vehicle Systems

Chamlmers University of Technology Gothemburg, Sweden

TRAFFIC CONGESTION AND ASSOCIATED COLLISIONS

Submitted by

Jacobo Antona Dakota Gale

Fransois Maurice Olivier Dorlot

Kietsakul Watcharinyanon Borja Alonso de Linaje

2

TABLE OF CONTENTS

1. Introduction ................................................................................................................................4 2. Accident data analysis ................................................................................................................4 3. Active safety systems to avoid crashes ......................................................................................7 3.1 GPS ..…………………………….……………………………..………………………….8

3.1.1. Introduction..............................................................................................................7 3.1.2. Description of Technology.......................................................................................8 3.1.3. Avaibility on the market ..........................................................................................8

3.1.3.1. GPS Advance Warning for Collision Avoidance...........................................8 3.1.3.2. Emergency Location ......................................................................................9 3.1.3.3. Future Developments of GPS.........................................................................9

3.2 Adaptive cruise control 3.2.1. Introduction..............................................................................................................9 3.2.2. Technology description..........................................................................................10 3.2.3. Adaptive cruise control in traffic congestion.........................................................10 3.2.4. Availability in the market ......................................................................................11

3.3 Brake assitant system 3.3.1. Introduction............................................................................................................11 3.3.2. Technology description..........................................................................................12 3.3.3. Brake assistant system in traffic congestion ..........................................................12 3.3.4. Availability on the market......................................................................................13

3.4 Lateral/side sensors and control systems 3.4.1. Introduction............................................................................................................13 3.4.2. Technology description..........................................................................................13

3.4.2.1. Lane Departure Warning System (LDWS) ..................................................13 3.4.2.1.1. Technology description ...........................................................................13 3.4.2.1.3. LDWS in a traffic congestion..................................................................15

3.4.2.2. Side Sensing : Blind Spot Monitoring and Lane Change Assistance ..........15 3.4.3. Evaluation in a traffic congestion ..........................................................................16

3.5 Computer vision for driving assistance 3.5.1. Introduction............................................................................................................17 3.5.2. CV in traffic congestion.........................................................................................17 3.5.3. Availability on the market......................................................................................18 3.5.4. Conclusion .............................................................................................................18

4. Traffic congestion avoidance ...................................................................................................19 4.1 Introduction 4.2 Different types of traffic simulation models

4.2.1. System to avoid traffic congestions in interurban roads ........................................19 4.2.1.1. Introduction ..................................................................................................19 4.2.1.2. Model technical survey ................................................................................20

4.2.2. System to avoid traffic congestions in urban roads ...............................................21 4.2.2.1. Introduction ..................................................................................................21 4.2.2.2. Model technical survey ................................................................................22

4.3 Queue theory 4.3.1. Introduction............................................................................................................23 4.3.2. Technical explanation about the model .................................................................23

5. Conclusion................................................................................................................................26 6. References…………………………………………………………………………………..30

3

TABLE OF FIGURES Figure 1, Congestion Pipe Figure 2, Number of injury accident, 2002 Figure 3, Number of fatalities, 2002 Figure 4, Relation between traffic volumes and crash rates Figure 5, ACP4 on-line system display Figure 6, SCOOT, DIME, HUTSIM scheme Figure 7, Queue theory scheme Figure 8, Queue discipline (FIFO) diagrams Figure 9, Single channel model for queue theory TABLE OF PICTURES Picture 1, Urban traffic in LA Picture 2, LDWS display Picture 3, Signal flow in ACC Picture 4, Autobahn Web Page traffic road map Picture 5, Reduction times using SCOOT in Beijhing Picture 6, Scoot Online Computer Office, UK Picture 7, Highway Toll System

4

1. Introduction

Road traffic has been increasing since cars were invented. The latest trends [1] show that, due to freight demands and increased personal vehicle use, it will get even worse . With the increasing number of cars and trucks on the road, traffic congestion will only increase. In addition to causing stress, wasting time, and decreasing mobility, congestion increases road crashes, vehicle operating costs, fuel consumption, and air pollution. This report discusses the problem of traffic congestion and collisions associated with congestion. Traffic congestion will be defined and the different types of accidents occurring in those situations will be identified. The purpose of this report is to find solutions to reduce the number of injured and killed people in these accidents. The development of the paper it is divided into two main parts: the first will explain the latest devices required on cars to prevent accidents in traffic congestion situations, and the second deals with the reduction of traffic congestion.

2. Accident data analysis

2.1. Traffic congestion

As traffic continues to increase on a relatively unchanging highway network, congestion has become part of daily life in many places.. Highway congestion is not just a problem of recurring "rush hour" delay in major cities. More than half of all congestion is non-recurring—caused by crashes, disabled vehicles, adverse weather, work zones, special events and other temporary disruptions to the highway transportation system.

Picture 1

The European Conference of Ministers of Transport [3] adopted the following definition of traffic congestion: “Congestion is the impedance vehicles impose on each other, due to speed-flow relationship, in conditions where the use of a transport system approaches its capacity.” Congestion is the “resistance” slowing traffic as the number of vehicles increases towards saturation. Simply, highway congestion is caused when traffic demand approaches or exceeds the available capacity of the highway system. While this is a simple concept, it is not constant. Traffic demands vary significantly depending on the season of the year, the day of the week, and even the time of day. Also, the capacity, often mistaken as constant, can change because of weather, work zones, or traffic incidents.

5

Figure 1,

Congestion pipe

Between 1980 and 1999, route miles of US highways increased 1.5 percent while vehicle miles of travel increased 76 percent, according to FHWA [1]. In 2000, the 75 largest metropolitan areas experienced 3.6 billion hours of delay, resulting in $67.5 billion in lost productivity, according to the Texas Transportation Institute.

2.2. Crashes in traffic congestion

A crash that occurs because of the congestion or distraction from a prior incident is referred to as a "secondary crash". Exact figures on the number of secondary crashes are difficult to calculate, however two studies estimate that around 15 percent of crashes are the result of an earlier incident. Secondary crashes often can be more serious then the original crash, especially if they occur at the boundary between freeflowing, highway speed traffic and stopped traffic. Furthermore, congestion increases crashes.. Because they are influenced by a variety of different factors, the system involving traffic congestions and these crashes is quite complex. It can include any of the following: · Human factor · Vehicle technology · Road infrastructure, · Public and private costs · Weather

6

Congestion facts :

• As could be predicted, the peak time for crashes coincide with rush-hour congestion.

• Congestion caused by a single crash contributes to subsequent crashes.

• Rear-end collisions comprise 44 percent of all beltway crashes. The two principal causes are the driver inattentiveness and differential speeds. In most cases (73 percent), the lead vehicle slowed or stopped because of traffic congestion.

• Approximately 17 percent of crashes occurred when one vehicle sideswiped another. This crash type is typically related to congestion or frequent lane changes.

Congestion crash rates are not what it might be expected. A recent study by researchers from the Michigan Department of Transportation and Michigan State University looked at a sixteen mile segment of Interstate in the Detroit area over two years to examine relationships between different levels of congestion, measured by volume to capacity (v/c) ratios, and crash rates, measured by crashes per 100 million vehicle miles travelled (VMT) [4]. The results indicated that crash rates are very high at low levels of congestion but rapidly decrease with increasing v/c ratios, before gradually increasing again at peak levels of congestion. This U-shaped model holds true for overall weekday and weekend crashes, multi-vehicle crashes, rear-end crashes, and property-damage-only crashes. On the other hand, injury and fatal crashes tend to decrease steadily as v/c ratios increase. Why are overall crash-rates higher when congestion is lowest? First, low traffic levels permit higher speeds. Single-car fixed-object and rollover crashes are more prevalent. Also, lower v/c ratios typically occur late at night and in the early morning, when drinking and drowsy drivers are prevalent. The authors attributed the increase in general crashes at very high v/c ratios to the increment of traffic conflicts. Interestingly, this increase appears to limit the number of fatal and injury crashes.While accidents are more likely to be fatal at higher speeds (Figure 3), fixing bottlenecks can nevertheless reduce the number of all types of crashes, thereby saving lives and preventing injuries. A study prepared for the American Highway Users Alliance estimated that improvements to the 166 most serious bottlenecks nationwide would prevent 287,200 crashes over a 20-year period, including 1,150 fatalities and 141,000 injuries.

Figure 2, Number of injury accident, 2002 Figure 3, Number of fatalities, 2002

7

This figure illustrates the relationship between traffic volumes and crash rates. Congestion increases as the volume-to-capacity ratio approaches or exceeds 1.0. A greater level of traffic volume relative to the capacity of the highway increases the risk of a crash.

Figure 4

The literature reviewed also identifies the standard deviation of speed to be an important characteristic that affects crash rate. Although research has related crash rates to traffic volumes and highway geometric characteristics, very little research has been done to relate number of crashes to the flow or occupancy. Garber and Ehrhart identified several models that relate crash rates to speed and flow characteristics.

3. Active safety systems to avoid crashes

3.1. GPS

3.1.1. Introduction

One potential solution to the problem of collisions between vehicles in a traffic congestion zone is the use of the Global Positioning System (GPS). A fast evolving technology, GPS holds much promise for advance notification in traffic congestion scenarios. As prices decrease, availability will increase as more companies implement cheap and effective methods of GPS usage in vehicles. This report section includes a description of GPS technology, its application as related to collision avoidance between vehicles, a brief description of evolving GPS technology, and a discussion of the economic feasibility of on-board vehicle GPS traffic collision avoidance systems.

8

3.1.2. Description of Technology

GPS technology is comprised of three parts: 1) Satellite broadcast signal; 2) Ground control stations; and 3) Independent GPS receivers throughout the world. The satellite system is a satellite network initially launched by the U.S. Department of Defense in 1978, but made available for public use in the 1980s. While one channel is reserved for military use, another channel is designated expressly for civilians. Operating on a continually broadcasted frequency of 1575.42 MHz, GPS works in any weather conditions, 24 hours a day anywhere in the world – perfect for traffic solutions because vehicles are always operating somewhere on the globe. The satellite network is a system of 27 solar-powered satellites (24 in operation and 3 backups). The satellites, orbitting the Earth at an altitude of 19,300 km, make two revolutions per day, and the spacing ensures that at least five satellites will be visible at any point on Earth at any time. Using a visual line-of-sight communication with the satellites and 3-D triangulation, the location of a GPS receiver can be determined. When first introduced, timing errors were included in GPS technology to limit non-military receiver accuracy to only 100 meters. This was changed in 2000, and a new technology known as Differential GPS (DGPS) now allows accuracy of 1 meter, extremely important for lane-level solutions to traffic. (1) DGPS works by removing the delay error associated with four moving points. By using a stationary receiver, the satellite measurements can be tied into a solid local reference, and by tracking the signal in reverse back to the satellite, any time delay can be compensated for and the resolution of the signal accuracy can be dramatically increased. (2)

3.1.3. Avaibility on the market

The use of GPS in cars started around 1995 and has continually evolved since that time. New technologies are constantly being developed for use in traffic collision avoidance, but a flawless solution has not yet come close to being achieved.

3.1.3.1. GPS Advance Warning for Collision Avoidance

The primary application of GPS technology is its implementation in advance warning systems. By detecting a potential hazard in the road ahead, such as slowing traffic, a GPS system can alert a driver in time to allow evasive maneuvering and hopefully avoid collision entirely. DGPS is especially useful for this, as the increased resolution and accuracy allows for greater predictive abilities for the system. Due to legal liabilities, car companies are hesitant to install systems where the car entirely takes over control of the vehicle (3). A warning of impending hazards, such as what DGPS is capable of, therefore has extreme potential. For drivers logging tens of thousands of kilometers per year, and especially those driving frequently driving in congested situations, an advance warning system can potentially reduce accidents by up to 50%. (4) Transport companies are especially impacted by this, as their drivers covers millions of miles per year and encounter thousands of traffic congestion situations per year. The system can reduce the number of accidents in a number of ways. First, it can direct the driver to entirely avoid a congested traffic situation, essentially operating as a high-tech radio traffic channel. By steering clear of the congested area, not only does the driver stay safe but their personal contribution to the congestion is avoided. Known as rubbernecking, congestion is often caused by curious drivers slowing down to look at accidents or anything out of the ordinary along a roadway.

9

This often causes additional traffic congestion on both sides of the road and can result in further accidents. (5) If enough cars are equipped with these types of systems, rubbernecking would not occur as often and additional congestion would be reduced. Emergency vehicles could also have easier access to accident scenes, saving lives in the process. Secondly, the system could be used in conjunction with signal lights, similar to the traffic-metering techniques encountered in many large cities in the United States. By anticipating the amount of traffic between a vehicle and its destination, a GPS unit can communicate with a signal system and not join the crush of cars approaching a congested situation all at once. This slower merging with major traffic flow can greatly decrease traffic congestion and reduce the possibility of accidents.

3.1.3.2. Emergency Location

GPS systems, in connection with car-mounted cellular telephones, are now being used to automatically locate cars involved in accidents. If it detects that the airbag has deployed, a microcomputer calls a service center over the car's cell phone while passing it the last known location of the car as determined by the GPS receiver. The service center passes the information to the local emergency services, who can then respond rapidly. This can help prevent future congestion and additional accidents by removing the rubbernecking attraction (6).

3.1.3.3. Future Developments of GPS

This field is currently evolving at a remarkable pace. A perusal of patents related to GPS traffic collision avoidance yields huge numbers recent new inventions, and it seems many companies are attempting to take advantage of the current possibilities of on-board accident prevention GPS technology. The increased resolution of DGPS has allowed for accuracy, and according to Professor Bradford Parkinson of Stanford, the number of GPS users will surpass 50 million worldwide (7). This increased dispersion will result in a higher number of systems on the market and a decreased cost for manufacturers. Also, increased user familiarity will result in a greater ease of implementation and therefore increased usage by drivers. GPS collision avoidance is currently being researched by all the major car companies: Ford, GM, Volvo, BMW, Mercedes – all are striving to provide easy and safe methods to harness this technology. The future will contain many new developments that will most certainly lead to a decreased accident rate saving many lives and millions of dollars.

3.2. Adaptive cruise control

3.2.1. Introduction

In traffic congestion, concentration and the ability to react to any road situation is required. Moreover, drivers must maintain the constant distance and relative speed of the vehicle in front by adjusting their own speed accordingly.

10

Other than accelerating and braking the vehicle, those task can be perform by the adaptive cruise control (ACC), which is the extension of the conventional cruise control system. In addition to keeping the vehicle at a set constant speed, ACC can be automatically slowed down and then follow the slower vehicle at a set distance when the vehicle is confronted with a slower moving vehicle ahead. Once the road is clear again, the ACC accelerates the vehicle back to the previous set cruising speed.

3.2.2. Technology description

The basic principal of this application is the use of forward-looking sensor installed behind the grill of the vehicle to detect the speed and the distance of the vehicle ahead it. When the slower vehicle in front or another object is detected by the sensor, digital signal processor and longitudinal controller will generated the signal to control the engine and braking system to reduce the speed of the vehicle. The figure below shows how the signal flows in the system.

Figure 5, Signal Flow in ACC[4]

One of the most important part of the system is the radar sensor. This part can be specified by the range and speed of vehicle it can detect by the radar. The most common type on the market now is the 77-GHz radar system made by TRW. It has a forward-looking range of up to 492 feet, and operates at vehicle speeds ranging from 18.6 miles per hour to 111 miles per hour.

3.2.3. Adaptive cruise control in traffic congestion.

One main cause of traffic congestion is the propagation of reduction in speed of vehicles on the road. Consider if one who drives a car leading the traffic queue slams on the brake, perhaps to let another car merge in to the lane; the driver behind will respond by braking too as does another one behind. All drivers behind trend to brake harder than the one in the front in order to be in the safest zone.

Normally people take some time to respond to speed changes. Such sluggish reaction times mean that extra space is needed between vehicles for safety.

11

ACC is one potential solution for this problem. Since the radar sensor and electric control system respond much more quickly and precisely than human driver in changing speed. A vehicle using ACC typically brakes sooner and more smoothly than one without it.

By the use of computer simulation, applying concepts from statistical mechanics, to study the effect of ACC on the traffic flow, the result show that if all vehicles on a highway had ACC, perturbations due to changes in the lead vehicle’s speed would not translate into propagating pockets of traffic congestion, no traffic jam would form. At the speed of 67 miles per hour, if only 20% of vehicles on the highway used ACC, the flow remains smooth without any traffic jam. At lower concentration of ACC used vehicle, intermittent of traffic congestion would occur.

3.2.4. Availability in the market

Like most of other safety applications for cars, ACC is mounted in the vehicle during the manufacturing process. Currently, there is no such a system that can be used as a universal additional option in the market.

Availability of the system is quite narrow, as it is included only in luxury car produced by few manufacturers.

The system itself is not totally a safety feature but most likely a comfort feature. In the car with ACC, collision avoidance can not be guaranteed by the system. As same as all other safety systems, accident avoidance depends on driver responsibility.

For the future trend of this system, development by adding more radar sensor on other part around the vehicle is considerably useful. More or less, the end of the development seems to be the presence of automatic vehicles which need no more than a destination decision from the driver.

3.3. Brake assistant system

3.3.1. Introduction

The more congested the traffic is, the greater the possibility of an emergency situation becomes. This system detects emergency situations and converts the driver’s reaction into a smooth, powerful response that minimizes the stopping distance and consequently gets rid of or reduces the severity of a collision.

The radar sensor used in adaptive cruise control detects the distance and speed of the vehicle in front and can also serve as a brake assist system. The brake assist system is always active regardless whether ACC is activated or not.

The working principal of the brake assistant system is the integration of existing safety and convenience systems which are anti-lock braking system(ABS), Electric stability program(ESP), hydraulic brake assistant(HBA) and adaptive cruise control(ACC). By combining and networking these systems, it is possible to recognize accident in advance and perform an effective response.

12

3.3.2. Technology description

The operation of the brake assistant system can be described by dividing it into three subsystems as explained below.

Predictive brake assistant (PBA)

After the sensor detect an accident risk, the predictive brake assistant (PBA), as the first stage of the brake assistant system, prepares the braking system for an emergency stop. Making the brake ready to perform the proper operation reduces the braking distance, PBA fills the brake circuit in advance, achieving the required pressure more quickly and positioning the brake lining close to the brake discs. All of this happens without the driver noticing. The system also reduces the tripping level of the hydraulic brake assistant, making it respond more quickly

As the above occurs, the driver gains valuable time. As soon as they step on the brake pedal, full braking performance is available, allowing shorter braking distances.

Predictive collision warning (PWC)

Second module of the brake assistant system, predictive collision warning (PWC), additionally warns the drivers of critical situations. A short burst on the brakes, a brief tug on the safety belt, and visual and acoustic signals alert the driver to the imminent danger.

The study by the Association of German Insurers shows that almost half of all drivers involved in road accidents did not brake at all. Early warning allows drivers to react faster to the danger of collision by taking evasive action and braking to reduce the impact speed.

Predictive emergency brake (PEB)

If the drivers, for any reason, do not react despite all warnings and the system recognizes an unavoidable accident due to the detection of position and speed of the other vehicle, the predictive emergency brake (PEB) actuates automatically. Until this moment, the drivers are merely supported by other after-crash safety systems like air bag and seat belt. But from then on, PEB activates and emergency braking maneouver at maximum strength uses the last chance to reduce the force of an unavoidable collision and minimize injuries.

3.3.3. Brake assistant system in traffic congestion

During accident risk situations, only a third of drivers respond with the proper braking action. In congested traffic, there is more possibility of collision from all direction due to less space between each car, and consequently less time for the drivers to make a decision. Moreover, high level of concentration is required for a long period, resulting in fatigued condition of the eyes, arms, legs and other parts of the body leading to slower reaction time.

13

To increase the possibility of avoiding or reducing the aggression of the collision caused by traffic congestion, brake assistant system is an effective solution. The system can convert the inappropriate reaction of the driver into proper braking operation.

3.3.4. Availability on the market

Since brake assistant system is the integration of other safety systems, it is not available as an additional option for the car without those systems.

The fully brake assistant system is not currently used on the road since the technology has been introduced for a few years. The first stage of the system, PBA was used on the real road and the second stage, PCW is almost ready for serial production.

As mentioned, the system has a lot of effect on the behaviour of the vehicle. There might be some other possible solutions before the system acts. The best solution may not create a fully automatic system so the development on how to make it flexible is considerably important. Further development should focus on how to make the system cover the largest variety of accident types.

3.4. Lateral/side sensing and control systems

3.4.1. Introduction

Traffic congestion simulations have shown that in heavy but free flowing traffic, congestions can arise spontaneously, triggered by minor events such as an abrupt steering manoeuvre by a single motorist. That is whyit is necessary to implement control systems assistance to assist the driver during some difficult manoeuvres due to the dense flux and help better control his behaviour which might concern to other drivers. Driver assistance systems are all being actively researched and already implemented into commercial production. It is therefore important to take a critical look at aspects of these systems, [1] namely the Lane Departure Warning System (LDWS) and Lane Change Assistance (LCA).

3.4.2. Technology description

3.4.2.1. Lane Departure Warning System (LDWS)

3.4.2.1.1. Technology description

The LDWS application [2] provides the driver with either an optical or audible warning (or combination of the two) when the vehicle crosses a lane. The system fits a three-parameter road model: lateral position, slope and curvature. The application can establish indications to unintentional roadway departure that can be a hazardous situation during congestion and therefore the application gives a warning to the driver.

14

Picture 2

The picture above illustrates the lane following by graphically overlaying the lane tangents (green lines), which take into account position and slope and the road skeleton (blue line) takes into account the curvature estimation. Note that it is relatively simple to manipulate this system thanks to the colors used.

3.4.2.1.2. Availability on the market

Lane Departure Warning System originally entered the heavy truck market in Europe in 2000, followed shortly thereafter by the United States. The systems became available to car drivers in Japan initially and were first introduced in Europe and the United States in 2004 on Infiniti and Citroën vehicles, respectively. The systems described below are representative of the various products on the market [1].

� Assist Ware SafeTRAC System : The SafeTRAC System is marketed to the North American heavy truck market. SafeTRAC is unique because provides a continuous indication to the driver of vehicle position within the lane, via a simple graphical display

� Iteris LDWS : The Iteris Autovue system is the market leader in LWDS. Originally introduced in Europe in 2000, over 8.000 units have since then been sold there and sales are averaging 4.000 systems annually. In the United states, Autovue is now available as a factory option from several truck makers and over 600 units have been sold. In the auto market, Iteris is also the supplier of the LDWS introduced by Infiniti and Citroën.

� MobilEye LDWS : MobilEye has pioneered the development of application for driver assistance. The system became available to automotive and truck fleets as an after market product in 2004.

� Toyota Rear-view system : The Toyota LDWS on the market in Japan takes an innovate approach in using the rear view camera for double duty. The Toyota system, developed cooperatively with supplier Aisin, uses the same camera to look at lane markings immediately behind the vehicle while on the highway to realize the LDWS function.

15

3.4.2.1.3. LDWS in a traffic congestion

The Dutch Ministry of Transport sponsored field trials of LDWS 2002 and 2003. The trials were focused on professional drivers operating heavy-duty and long-distance buses. Overall, the effect of LWDS on traffic safety has proved to be positive. The results indicated that, with all trucks in the Netherlands equipped, approximately 10% of injury crashes involving heavy vehicles could be prevented. With respect to traffic flow, LWDS are not expected to have either a positive or negative influence, other than the reduction in congestion due to fewer truck crashes. 75% of the drivers had positive opinions of LDWS, and over 50% stated that they would prefer to drive, in particular during traffic jams, with such a system installed in their vehicle. However, 21% of drivers stated that they would prefer vehicles without LDWS. On the positive side, 60% of drivers concluded that the system caused them to pay more attention to the driving task. On the down side, 20% of the drivers felt that the system could cause a frighten response that might be worse than crossing the lane line.

3.4.2.2. Side Sensing : Blind Spot Monitoring and Lane Change Assistance (LCA)

3.4.2.2.1. Technology description

Side-sensing helps drivers make safe lane changes on a motorway in detecting vehicles within the "blind spot" area. The lane change assistance detects vehicles in adjacent lanes which may be rapidly approaching and could also put a hazard in a lane change manoeuvre. Being a simpler problem, blind spot monitoring systems are more mature and basic systems are on the market. The primary sensing modality is short-range radar. Vision-based systems for both blind spot monitoring and LCA have also been implemented.

3.4.2.2.2. Availability on the market :

The pioneer in blind spot monitoring is the Eaton VORAD system, which was introduced to the heavy truck market in the early 1990s. The VORAD system uses a radar at 24 GHz to help drivers to detect vehicles in the right side blind spot of the typical large truck. The sensor is pointed at a right angle to the truck and covers the adjacent lane. For the automotive market, the challenge is to offer radar-based sensing systems at a much lower cost than the systems sold to trucks. A brief insight of systems from Valeo and Visteon is given next [1].

� The Valeo system uses 24 GHz radar sensors to monitor the blind spot on both sides of the vehicle. If a vehicle appears in the blind spot, the system alerts the driver through a visible icon. The systems range extends to 40 meter, with a 150 degrees broad field view. The sensors are integrated into the vehicle behind the rear plastic bumper. Valeo expects its system to appear on production vehicles around 2006.

� The Visteon system is a close to blind spot monitor, covering a detection distance of 6 meters, again using 24 GHz radar. Its driver interface approach uses an illuminated icon integrated into the side mirror. To avoid bothering alerts, the system is designed to delay an alarm for what appears to be an overtaking vehicle that is likely not a threat. If the approaching vehicle does not pass the host vehicle, so no alarm is sounded at all.

16

3.4.2.2.3. Technology description

MobilEye offers blind spot monitoring and lane change assistance using a single camera solution [3]. By mounting a camera on the side mirror, we can use for monitoring the host vehicle's blind spot area and also detect approaching vehicles from a distance of 60m. The combined analysis of the near and far areas as viewed from the side camera is geared towards indicating to the driver (by means of a Red/Green Led, for example) whether it is safe to change lane or not. The blind spot area analysis is based primarily an outward flow of pixels in a normal situation. A violation of this assumption may indicate the presence of a foreign object in the blind spot area.

Picture 3

The above picture is taken from the ACP4 on-line system response. The Orange rectangle signals the area of blind spot analysis, vehicles detected by the system are marked in Green and vehicles posing a threat (based on distance and time to contact) are marked in Red. When a vehicle (or any other object for that matter) is in the blind spot area the text "BSD" appears in the Yellow area. Finally, the Red/Green box in the upper right corner indicates what the driver would see as an output of the system, Red meaning it is unsafe to change lanes and Green indicating safe. Consider a multi-lane situation. The system detects the lane position of the approaching vehicles. Vehicles that are one lane removed are marked by a blue rectangle and do not affect the Red/Green indicator. The system performs well under all conceivable driving conditions and is invariant to traffic density, performig equally well in congested and open highway conditions.

3.4.3. Evaluation in a traffic congestion

From full implementation of Intelligent Transport Systems (ITS) in United States [4], the lane change crash avoidance systems can reduce lane change accidents by 37%. ITS also estimates that there were 282 fatal crashes of the type sideswipe/same direction in 1995. Similarly, they use sideswipe/same direction crashes to estimate the target crash size for lane change. Of the 282 fatal crashes, they estimate that infrastructure-based and cooperative ITS countermeasures will have already reduced 78 of them. Therefore, in order to avoid double counting, they adjust their sideswipe/same direction crash type size to be 204 fatal crashes. Using the 37% reduction rate, they finally estimate a reduction of 75% fatal crashes. Of course, among these 75% avoided fatal crashes, we can estimate that a lot of avoided crash could come from during congestion because it is in this case that the lane change system is more used.

17

3.5. Computer vision for driving assistance

3.5.1. Introduction

Traffic congestion can generate a phenomenon of nervous tiredness due to a dense and a slow circulation. What is more, the fatigue phenomenon slows reaction time, decreases awareness, and impairs judgment. Therefore, a driving assistance with the computer vision can help the driver avoid or limit crashes due to fatigue which may play a role in behaviour, namely lapses of attention. Similarly, during the situation of a congestion, the driver has to endure long hours of patience. This can lead them to read the newspaper, put on makeup, or concentrate on a cell phone conversation instead of focusing on the road and driving. Those are the most obvious examples of driver distraction. Moreover, a research indicates that driver distraction is a contributing factor in 25-50% of all crashes. That is why the computer vision for driving assistance can play a major role so as to avoid collisions and reduce accidents.

3.5.2. CV in traffic congestion

Three main reasons promote the development of computer vision systems for cars [5]. Regarding safety

Further considerable progress will be possible with sensor systems that perceive the environment around the car and are able to recognize dangerous situations. For instance, they will alert the driver if he is leaving the lane, disregarding traffic signs and lights, or overlooking a possible collision, all important factors in traffic congestion. The important advantage of vision-based systems is their potential to understand the current traffic situation, a prerequisite for driver warning or interventions in complex situation, such as the dense circulation during congestion. Vision-based driver assistance systems allows an unprecedented increase in driving convenience. Tedious tasks like driving in stop-and-go traffic due to congestion can be taken over by the system as well as distance or lateral control on highways. Regarding its efficiency

It is obvious that less traffic accidents mean less traffic jams and less economical loss. In addition, computer vision can be used to automate traffic on special roads or to improve the efficiency of goods transport by coupling trucks by means of an electronic tow-bar system and thereby possibly avoiding accidents during traffic jams.

18

3.5.3. Availability on the market

Technology is rapidly increasing the capabilities of modern vehicles. One of the most significant development for the next decades is the emergence of a set of vehicle capabilities centred around the notion of driver assistance. These involve sensor-based systems, which continuously evaluate the surroundings of the vehicle, display relevant information to the driver and might even take control of the vehicle. As such, these driver assistance systems hold great promise in increasing the safety, convenience and efficiency of driving.

Safety systems start by offering such basic capabilities as increased visibility in bad weather or at night. Cadillac DeVille's IR-based night vision system, for instance, projects an image of the road ahead on a small patch of the windshield. The Lane Departure Warning system, for example, available in Mercedes-Benz' and Freightliner's line of heavy trucks, involves a video camera looking forward on the road. The system produces a rumbling sound if the vehicle changes lanes without the prior use of a blinker that is, in such case, a behaviour very hazardous in a congested situation. It is meant to alert those drivers which are swerving off the lane, possibly because they have fallen asleep, which is one of main risk for drivers during a traffic jam.

Collision danger lurks from the side and in front of the vehicle. Blind spot and lane change warning systems offer assistance in detecting approaching or overtaking vehicles. A number of Japanese vehicle manufacturers (e.g. Nissan, Mitsubishi) have announced the near-term availability of such collision warning systems for passenger cars.

Driver assistance systems also promise to increase convenience and comfort by relieving the driver from tedious parts of driving which is often caused by congestion. Indeed, congestion generates a fatigue due to stressful situation of monotony. Conventional cruise controls, which maintain a preset speed, appeared more than twenty years ago. As discussed earlier, adaptive cruise controls are perhaps the most visible component of sensor-based driver assistance systems nowadays. They are available as an option on most premium vehicles (e.g. Mercedes-Benz S-Class, Jaguar XKR and Lexus LS430). By 2002, Fiat plans to make it available in its mid-range Punto model.

3.5.4. Conclusion

A variety of sensors vie to provide the eyes and ears of driver assistance systems: video cameras, radar sensors (77 GHz / 24 GHz), laser scanners, and ultrasound devices. The human visual perception system is a good example of what performance might be achieved with such sensors, if only the appropriate processing is added

Apart from the technical challenges of developing driver assistance systems, questions remain concerning how they will affect the driver. Will the driver still be alert enough to intervene and take over control from a semiautomatic system when is required ? Will he drive more recklessly because he comes to rely on the driver assistance systems? All these issues remain to be examined, case by case.

To conclude, a fully intelligent vehicle must work cooperatively with the driver. Uncoordinated technologies could deliver excessive, competing, or contradictory messages and demands that might distract and overwhelm the driver.

19

4. Traffic congestion avoidance

4.1. Introduction

Traffic in big cities has become a major problem for the governments. It affects the overall efficiency of the city, the economy, the pollution, the quality of life and the traffic safety. Its reduction has become an important task for politicians and traffic experts all around the world. Traffic Simulation is an important tool for modelling the traffic behaviour and for analyzing the causes and potential solutions of traffic problems, including congestion and traffic safety.

4.2. Different types of traffic simulation models

The level of detail in simulation models range from macroscopic to microscopic. Macroscopic models describe the traffic at a high level of aggregation as flow (the number of vehicles per hour that pass a certain point) without considering its constituent parts, the vehicles. The microscopic models describe the behaviour of the entities as well as their interaction. Mesoscopic models are between the two other levels of detail, considering the individual vehicles but not their interactions. Microscopic simulation models provide a detailed overview of the traffic process, making these models more suitable. On the other hand, macroscopic and mesoscopic models have less detail of the traffic process but are faster and easier to apply. Mesoscopic models are ideal for prediction applications where the detailed modelling of driver interaction with the road network is not needed. Mesoscopic models are the base for traffic active systems, which are in use in many cities and countries all around the world. Some of them have been developed and used for both off-line and on-line prediction of traffic conditions.

4.2.1. System to avoid traffic congestions in interurban roads

4.2.1.1. Introduction In 1992 Michael Schreckenberg (University of Duisburg-Essen) and Kai Angel (University of Berlin) developed a mathematical model able to predict traffic jams. This simple mesoscopic model accurately simulates congestion phenomena in motorways and traffic waves. Many of simulation systems have been developed based on this system. A real system based on this model is already working in the motorway network in North Rhein Westphalia (Germany) and it is able to predict during 90% on-time situation. Free access for the public is available on the official webpage, http://www.autobahn.nrw.de , where every driver can check the traffic situation and predictions within a reasonable period of time. .

In this system, the simulation model is continuously fed and complemented by real traffic data (2500 automatic data detectors send information about vehicle speed and traffic flow to the system every minute) However, unexpected problems have been found. Three hundred thousand drivers per day are checking the information and changing their routes depending on the predictions, strongly affecting the efficiency of the system. The traffic is also affected by many disturbances, and collecting them is a challenge.

20

Picture 4

Scientists agree that one of the best points in this system is that human behaviour is introduced until certain level in the model. Difference between `defensive´ and `aggressive´ drivers can be set as an input in the simulation model. The model includes gradual decelerations close to real life. A handicap of this kind of systems is economical cost. A digital road map with a detailed representation of the freeway topology, including tangents, ramps, acceleration lanes etc, is the basis of this simulation software and the initial system set-up cost can be exorbitant

4.2.1.2. Model technical survey

This model is based on the cellular automata (dynamic system that evolve by discrete steps); a mathematical model that includes physics and statistics concepts. Every road is divided into cells, one cell for each individual car. The number of empty cells between one car and the next one depends on the type of driver. PARAMETERS used are position x (one for each car), velocity v (each vehicle has a single velocity, measured in number of cells advanced per time unit), vmx (maximum speed that can be reached by the vehicles, the same for all of them), probability p of a vehicle reducing its speed randomly (a distraction of the driver before accelerating from stop can be simulated here), and space b (number of cells between consecutive vehicles).

RULES. The behaviour of any car in the road can be modelled with 4 rules

R1.- Acceleration. V= min(v+1 , vmax) . Accelerate until the max velocity is reached R2.-Breaking because of the interaction with other vehicles v= min(v, b). The velocity will be the minimum between the one calculated in R1 and the space b to its predecessor car.

R3.- Random breaking with probability p. v= max(v-1, 0) R4.- Movement. x=x+v . The position of the car is updated based on the calculated speed

21

These are the parameters and basic rules used to simulate congested highways, with homogenic vehicles and only one lane. But to understand the basis of the more complex models, this can be enough. Multilane roads, accidents, working, different weather conditions, information feedback, crosses, braking lights effect, etc, are some of the parameters currently being implemented with more complex mathematical models, although most of them are based on this basic theory.

4.2.2. System to avoid traffic congestions in urban roads

4.2.2.1. Introduction

SCOOT, Split Cycle Offset Optimization Technique,[15] is a signalized urban network traffic control system developed in the United Kingdom. This system is based on a microscopic simulation model (HUTSIM). Detectors registering intersection movements in the network provide the information needed by the system to estimate arrival patterns to the traffic signals.

This system is implemented in over 200 cities in the United Kingdom (ROMANSE project)[16] and other countries. Successful results have been obtained with these systems with reductions over 10 per cent in the time used for the same way. Its implementation in the US is proving difficult because of infrastructures reasons, but nowadays it is working in many important cities. The next figure shows the results obtained by SCOOT in Beijing, where it has been active for the past 3 years.

Picture 5

The first attempts to coordinate and optimize signal settings were effective, but based on highly source-consuming complex plans for each time of the day with the necessity of being continuously updated.

To solve these problems, a methodology was developed. On-line computers continuously monitor traffic flows over the road net and make a series of frequent small adjustments to signal timings in order to reduce vehicle delays and improve traffic flow.

22

Picture 6

4.2.2.2. Model technical survey

The basis of the software is the same for all the SCOOT systems. The part of the software that is linking to the street equipment is specific for each supplier and city. There is continuous information feedback in both directions between the SCOOT and the road signal equipment. The software works with three parameters that can optimize the traffic signal settings:

Green Split (basically, the busiest stage gets the most green time), Cycle time (varies according to total vehicle demand through a group of junctions in a region. The busier the traffic, the higher this time should be), and Offset (to give progression to the biggest traffic flow. This keeps as much quantity of vehicles moving at any single time)

SCOOT gets traffic flows information from the detectors. Detectors are required in every link and its location is critical. It is possible to find studies where the detectors used were not designed for this application and they show excellent results when adapted to the SCOOT system. These alternatives are being studied specially in the USA due to the non-compatibilty of the current detectionsystems implemented for other purposes and the one required by the SCOOT System [17]. The system translates the signal of every vehicle passing into `link profile units´, a connection between flow and occupancy. `Cyclic flow profiles´ of lpu´s over time are built for each link. With this organized data, the information is sent to the traffic signals via control unit. The DIME, developed at NTU is Distributed Memory Enviroment that enables several pieces of the software to execute on net computers

Figure 6

23

4.3. Queue theory

4.3.1. Introduction

Different models are created in order to study and solve other traffic situations and their problems. In many situations, delays are produced by congestions created in the highway tolls, parking accesses etc. Queues created in these places can produce risk situations for the drivers, increasing the risk of back-rear collisions.

Picture 7

Queue sizes are important in the design of fixed facilities and in traffic management decisions. Usually, some compromise between size, waiting time and the cost of service must be made. Queues can be classified according to the number of available channels, however most queuing situations are single channel queues. The prediction of these queues and the consequent facilities design can remove part of the risk improving the driver security.

4.3.2. Technical explanation about the model

To model queue behaviour, it is necessary to specify some parameters of the system: [18]

Figure 7

Arrivals

λ= Mean arrival Rate 1/ λ = Time between arrivals The reality resembles a random process so a negative exponential distribution fits better.

24

Departures

µ= Mean departure rate 1/µ= Time between departures Although departure characteristics can fit with a random behaviour, is not very far from reality to consider a constant medium time.

Queues Discipline

Queues discipline defines which is the next unity acceding into the system. The most common discipline is FIFO (First In First Out).

Figure 8

The figures above represent the arrivals and departures. In the second one it is represented the overall process. LL(t) = Arrivals function S(t) = Departures function S*(t) = Departure function for vehicles that leaves the queue and enter into the service. N(t) = Number of vehicles in the queue at one single moment D(t) = Time spent by each vehicle according to FIFO

25

The area between the two lines represent the overall time spent by all the vehicles in the system. Single Channel model Considering the arrival and the service time exponentially distributed ,the figure 4 is obtained.

Figure 9

Some interesting parameters are defined: Probability of having n vehicles in the system:

( ) (1 )nP n ρ ρ= ⋅ −

Mean number of vehicles in the system:

2( ) ( ) (1 )

(1 ) ( )N s n P n

ρ λρ

ρ µ λ= ⋅ = − ⋅ =

− −∑

Mean time spent in the system

1/

( )ms sE N λ

µ λ= =

−

The mathematics of multi-channel queuing systems is more complex than for single channel queues. However, these multi-channel situations are more frequent in real world.

4.4. Conclusion

Many traffic management systems are being implemented nowadays all around he world. Its efficiency is still not accurately evaluated because there are many external factors that affect these systems. The experience of countries such as United Kingdom, Germany, China and USA should be used to improve the correct implementation of the Traffic Management Systems. These systems should be combined with correct traffic politics and vehicles laws in order to improve the overall current traffic situation.

Length

Time

26

5. Discussion About traffic congestions avoidance :

Traffic congestions affect to the overall efficiency of a country, to the pollution, to the quality of life of the citizenships and of course to the safety. The big quantity of things affected by the reduction of these traffic congestions send most of the responsablity to the governments, which always means completely different way to focus the problem compared of how private companies solve problems. The technical knowledge and the powerful of current technology could be enough to have much more effective systems to avoid traffic congestions all over the world; the biggest limitation is economical. Systems to simulate traffic conditions or to improve the road traffic need huge human and economical resources, which is usually a problem when no business is directly related. Once the decision of fighting against this is taken, unlimited number of systems can be developed to improve conditions. Few of the systems and models that are in use or being researched have been discussed. For the close future, it looks that the mathematical models already in use in several countries and the fast evolution of the capacity of information feedback between everything involved in a traffic congestion (weather information, vehicles, satellites, any communication system, general road information, statistical data base…) are a good combination to be optimistic in the development of this field. About the systems to avoid collisions in traffic congestions :

Governments have many things to say when thinking about avoiding accidents in traffic congestions because thousands of citizenships are killed every year all over the world. Specially governments can make new laws and give money for research about new safety systems, but the explanation of the fast evolution of the active safety systems developed for this purpose is based on the race started by companies a few years ago to be the safety. There is a big difference between how companies face the active safety fact, but the goal of all of them goes in the same direction, reducing accidents using systems economically feasible. Nowadays active safety is the field in the car industry where more human and economical effort is put due to the fast technical evolution and the importance of the safety in the cars we drive today. Two anecdotes:

Since the idea of the airbag arose until it became reality, 25 years were needed. No car in a few time will be built whithout airbag. When the idea of installing a GPS in a car was proposed some years ago in a meeting in Nissan company in Japan, some of the colleagues laughed at the group of engineers that made the proposal because it had absolutely no sense to implement such en expensive system when the main goal of the company was reducing costs. No car in a few time will be built whithout GPS or a system with the same functionality.

27

A short discusion about some of the systems we have researched:

GPS is fast becoming a more viable option for collision avoidance, especially with respect to traffic congestion. By evaluating potential hazards such as upcoming traffic jams, a system can help a driver entirely avoid potential accident zones. With the introduction of Differential GPS technology, the accuracy necessary to pinpoint individual lanes on a road is now available to the independent consumer. As the systems become more widely available, the price will continue to decrease and more manufacturers will install the systems in their vehicles. This opens the door to many new technologies related to control of the actual car braking and steering systems, allowing individual accident avoidance in case traffic congestion cannot be avoided. The next few years will reveal many new developments using GPS technology that will hopefully yield safer and more efficient transportation for everyone. Lane Departure Warning System (LDWS). Congestions can arise spontaneously, triggered by minor events such as an abrupt steering manoeuvre. It is necessary to implement control systems assistance so as to assist the driver during some difficult maneuvers due to the dense flux and contribute to control more his behavior compared to other drivers. So this system is relatively simple to manipulate thank to simple display video and graphically/cleverly used colours. Lane Change Assistance (LCA). We meet some area blind spot on whole vehicles which is one of more hazardous risk of collision. The risk is amplified during congestion due to heavy traffic. Therefore, the LCA is indispensable for the safety of occupant of a vehicle because it supports drivers. The LCA is also simple to manipulate. The driver is warned by simple display video if a vehicle rapidly approaches and could also put a hazard in a lane change manoeuvre. These two previous systems are really useful because they can reduce considerably the collision during congestion thanks to in particular on the fact that they are easy to use. Both of these systems are economically feasible and effective nowadays, so that is why the automobile industry implements more and more these systems on their vehicles.

Computer vision for driving assistance. Congestion can generate a phenomenon of nervous tiredness. What is more, the fatigue phenomenon slows your reaction time, decreases awareness, impairs judgment. Therefore, a driving assistance with the computer vision can assist the driver so as to avoid or limit crashes due to the fatigue which may play a role in behaviour namely lapses of attention.

Similarly, during the situation of congestion, the driver has to face up to long hour of patience. So the driver is subject to be distracted and a research indicates that driver distraction is a contributing factor in 25-50% of all crashes. So the computer vision for driving assistance can play a major role so as to avoid collisions and reduce accidents in particular during congestion. This system is in our opinion less necessary than LCA and LDWS because it mainly allows to coordinate several system of safety and it can tend to distract the driver or the driver may be less attentive because he will be able to say himself: "it is ok, I have a computer vision for driving assistance". Some things still have to change to make it succesful. Adaptive cruise control. As there is an announcement form the manufacturer of this technology that it is comfort feature not safety device, the adaptive cruise control can not be totally considered as a system that help the driver to avoid any collision. Moreover, using this system could lead to too-relaxing driving condition which might increase the risk of accident due to less intention of the driver.

28

The utility of the system is also limited. It can work only on highway speed without the intersection or pedestrian crossing line. In the congested traffic condition like city traffic, manual driving would be more effective in keeping the appropriate speed and distance from the car in front. Brake assistant system. The mistake of the driver due to less calm and purposeful response is one of the biggest reasons in car accident. The brake assistant system seems likely the direct solution which corrects the wrong action to the optimized performance to avoid or relieve the damage. However, with lot of variations of collision which varies in speed, position and dimension of the car, it is not able to cover all of them with only one or few assisted performances. Although the system can provide the function cover from additional braking to fully automatic braking, the other control systems of the car such as steering need to be managed to obtain the most appropriate result. Too-strong unpredictable corrective action from the system might causes worse result than the response of the driver without and assistant. Sensors are based on high technological breakthrough as it requires a lot of electronic and now appliance to be fitted on cars. Then it increases the overall price of the car when all this devices equip cars. Some equipment requires people adaptation for it to be efficient such as the sound alarm of laser radar. You have to know what beep speed corresponds to what distance. It is the same for the force feedback pedal. One has to get use to new equipments.

Integration of active and passive systems

By the correct integration of active and passive systems new levels of automotive safety would be achieve. Future cars will combine electronic systems that will network both active and passive vehicle safety systems and will provide support to the driver when responsive actions will be required. However, safety features such as seat-belts and airbags will continue to play a vital role in the car occupants safe. Nowadays the design of each single safety element should be thought as part of an overall safety system.According to this philosophy is possible to affirm that the total is more safety than the sum of its parts working as single elements. The potencial reduction of accidents with the integration of active and passive systems is big enough to continue the research in this field.The results will not be able in few years because it will take some time to include all this new systems in a significant proportion of the automotive park

29

6. Conclusion The proposed topic has been treated from its very general aspects until the most specific technologies that can be apllied to solve it. The first part of the report define traffic congestion and contain a general survey of traffic congestion stadistics.With those figures, the importance of solving traffic congestion and its associated collision its defenetly clear and remarkable. The study have reported many technologies that can contribute to the reduction of traffic congestion collisions, such as GPS, ACC, Brake Assistance Systems, LDWS, Side Sensing…however not all the technologies have the same grade of development.The reports disscusion describe the advantages and disadvantages of each technology and the necesary balance between the competition of private companies and governamental aids. Moreover, if the main proposa lis reducing traffic congestion collision, it is reasonable to focus the problem from another point of view, try to reduce the traffic congestion.Studies and appplied systems are being developed nowadays.Many cities all around the world use traffic management systems based on mathematical models that are improving traffic situation and therefore reducing collisions. A general overview of the problem is necessary. Goverments traffic politics and companies strategies should converge into the same direction to solve this increasing problem. Reduce traffic congestions and reduce its associate collisions should become one of the priorities for both of the two major agents in the system, governments and car companies.

30

REFERENCES

[0] Congestion charts trends US http://www.fhwa.dot.gov/policy/ohim/hs02/chartv.htm [1] European conference of minister of transport. http://www.cemt.org/online/conclus/rt110f.pdf [2] Zhou, Min and Virginia P. Sisiopiku, “On the Relationship Between Volume to Capacity Ratios and Accident Rates,” Presented at the TRB Annual Meeting January 12-16, 1997.

[3] http://www.nasm.si.edu/galleries/gps/work.html [4] http://www.trimble.com/gps/how1.html [5] Speech by Volvo guy on sensors [6] http://advanceautoparts.com/english/youcan/html/dsm/dsm20030801ap.html [7] Wikipedia, http://en.wikipedia.org/wiki/Motor-vehicle_collision#Rubbernecking [8] Intelligent Transportation Systems, US Departmnet of Transportation, http://www.its.dot.gov/gps.html [9] Lecture by Bradford Parkinson, Stanford University http://www.its.umn.edu/seminars/2003/3parkinson.html

[10] Intelligent Vehicle Technology and Trends, Richard Bishop, www.intelligenthighway.com/ITS/ivttsample.pdf. [11] MobilEye, www.mobileye.com/forwardLooking.shtml

[12] MobilEye, www.mobileye.com/sideMirror.shtml

[13] Estimating the potential safety benefits of Intelligent Transport Systems, Benjamin McKeever,

Nov 1998, www.itsdocs.fhwa.dot.gov//JPODOCS/REPTS_TE/6R01!.PDF

[14] From door to door - principles and applications of computer vision for driver assistant

systems, Uwe Franke, Dariu Gavrila, Axel Gern, Steffen Görzig, Reinhart Janssen, Frank Paetzold

and Christian Wöhler, DaimlerChrysler AG, www.gavrila.net/door2door01.pdf

[15] http://www.scoot-utc.com [16] http://www.romanse.org.uk/. [17] Performance Study of SCOOT Traffic Control System with Non-Ideal Detectorization: Field

31

Operational Test in the City of Anaheim UCI-ITS-WP-00-27

[18] GIVET,Departamento Ingenieria Mecanica ETSIIM ,Ingenieria del transporte, classnotes.

[19] FHWA. FHWA Congestion Mitigation website. http://www.fhwa.dot.gov/congestion/ congest2.htm. [20] Shelby Tedesco, Vassili Alexiadis, William Loudon, Richard Margiotta, and David Skinner,

Development of a Model to Assess the Safety Impacts of Implementing IVHS User Services, Proceedings, IVHS America, 1994.

[21] Davis, L.C. 2004. Effect of adaptive cruise control systems on traffic flow. Physical Review E

[22] http://www.bentleypublishers.com

[23] http://autorepair.about.com

[24] http://auto.howstuffworks.com

[25] http://www.bosch.ru/content/language2/html/715_2127.htm

[26] Intelligent vehicles, Prof. Dr. Dariu M. Gavrila,

www.gavrila.net/Computer_Vision/Smart_Vehicles/smart_vehicles.html

[27] Principles of Urban Transport Systems Planning. B.G. Hutchinson Scripta Book Company

McGraw-Hill Book Company

[28] http://www.autobahn.nrw.de [29] http://es.wikipedia.org/wiki/Modelo_Nagel-Schreckenberg [30] http://www.infra.kth.se/ctr/ [31] http://www.its.uci.edu/its/publications/papers/WP-00-27.pdf (scoot)