Dr. Tom V. Mathew


My major research focus inclues Traffic flow modeling and simulation and Transportation network optimization, control and management.


International collaborative research

Robust Network Design (2002-05)

With Satish V. Ukkusuri and S. Travis Waller of Civil Engineering Department-TRAN, University of Texas at Austin

This research addresses the problem of traffic network design problem (NDP) under demand uncertainty. The origin-destination trip matrices are taken as random variables with known probability distributions. Instead of finding optimal network design solutions for a given future scenario, we are concerned with solutions which are in some sense .good. for a variety of demand realizations. We propose a general definition of robustness and develop a methodology using genetic algorithm (GA) to find robust network design (RNDP) solutions. The methodology generates globally feasible near optimal network design solutions. The study makes two important contributions to the network design literature. First, robust network design solutions are significantly different from the deterministic NDPs and not accounting for them could underestimate the network wide impacts. Second, we systematically evaluate the computational performance by using different test networks and GA control parameter settings to obtain the appropriate parameters for this problem.

Transit Asset Management (2004-)

Snehamay Khasnabis, Professor of Civil Engineering, Wayne State University, Detroit, USA.

State Departments of Transportation (DOT) that provide the bulk of matching funds to local transit agencies for the purchase of new buses, are duly concerned about the escalating costs of new buses and the lack of sufficient funds to keep up with their replacement costs. This research tries to develop an asset management strategy for state DOT.s to meet their long-term fleet needs. Earlier models developed a two-stage process to allocate capital dollars for the dual purpose of purchasing new buses and rebuilding existing buses, when the needs of all the constituent agencies in a peer group are considered, and to distribute funds among the agencies in an equitable manner. The proposed management strategy includes two optimization models. Model 1 attempts to maximize the weighted fleet life of all the buses that are being purchased and rebuilt for a given peer group, within the constraints of a fixed budget. Model 2 is designed to maximize the Remaining Life (RL) of the entire peer group comprising the existing buses as well as those being replaced or rebuilt. The need for every single bus that is eligible for replacement is addressed in the model either through replacement or rebuilding. This two stage model is tried to combine into a single optimization model. A comprehensive case study depicting a seven-year planning cycle for the entire fleet of medium sized buses in the state of Michigan was used to demonstrate the strategy.

Robust Network Design II (2007-)

Satish V. Ukkusuri, Assistant Professor, RPI, USA.

This research first addresses the computational issue of Robust Network Design (RND) by adopting improved sampling techniques. Futher, the network reliability concepts need to be incorporated.

Doctoral Research

Urban Bus Transit Route Network Design Using Genetic Algorithm (1994-1999)

Tom V Mathew

With S B Pattnaik Professor (Rtd) and S Mohan Professor, IIT Madras

In the thesis a Genetic Algorithm based model was developed for designing an urban bus transit route network. The model was studied in different networks and applied to a large case study network. The size of the network is largest reported network in terms of number of nodes and links. The model behaved consistantly in all the networks. (Publications: Pattnaik et al. (1998)Tom & Mohan (2003))

Heterogeneous Traffic flow modeling and simulation using cellular automata (2002-05)

Pradip Jayantilal Gundaliya

with SL Dhingra

Arterials form the backbone of any rural and urban road system. In developing countries, the road network system serve heterogeneous traffic with a wide range of variations in driver behaviour and static and dynamic characteristics of vehicles. In the present scenario, vehicular traffic is increasing at an exponential rate which has created problems like congestion, parking, environmental pollution, etc. Moreover, limited road space in urban areas restricts mobility and comfort. In turn, there has been an increased interest in traffic flow modelling which represents the vehicular traffic behaviour and gives relations between speed-flow-density. Research on traffic flow modeling started about fifty years ago and has resulted in serveral models describing different aspects of traffic flow operations, either by considering the influence of vehicles individually (microscopic models) or from the viewpoint of collective vehicular flow (macroscopic models).

Most of the traffic flow models are developed for homogeneous traffic where majority of the vehicles are passenger cars. These models cannot be implemented directly for the areterials of developing countries having vehicle with diverse physical characteristics. Due to the complexities involved in developing mathematical models for heterogeneous traffic, simulation based approaches are usually adopted by most researchers. Car following theory based on Newtonian laws are used to describe movement of vehicles in these models. In this approach the postion, speed, and acceleration of every vehicle at each time interval need to be computed from basic equtations of motion. Althogh this approch provides accurate individual vehicle details, they require high computational ressources making it unsuitable for many real time application. This has motivated the developement of computationally efficient models with reasonable accuracy. In this context, cellular automata (CA) has emerged as an efficient tool in modelling traffic flow. CA discretizes continuous variables and uses simple vehicle movement rules (acceleration, deceleration, randomisation, and updation) to describe traffic flow. Several studies demonstrated the use of CA in traffic flow modelling and proved its computational efficiency.

In this study, heterogeneous traffic flow modeling using cellular automata is attempted for a urban arterial. In this approach, CA model for homogeneous traffic is modified for heterogeneous traffic conditions. Accordingly, three models are proposed: single lane traffic flow model (SLTF), two lane traffic flow model (TLTF), and grid based traffic flow model (GBTF).

The first model (SLTF) is developed based on model developed by Nagel and Schreckenberg (1992) for freeway known as NaSch model. To incorporate heterogeneity, the NaSch model is modified by changing cell size based on the vehicle types and by considering different types of vehicle depending on their dynamic characteristics like maximum speed, acceleration and deceleration. It is also proposed to use modified randomisation rule for better performance of the model in heterogeneous traffic flow. In addition to this an incident occurrence rule is developed to study the behaviour of vehicular movement at the incident place. The model requires information like cell size, which represent the standard vehicle with the required clearance; types of vehicles and maximum velocity of each type of vehicle; the initial density (cells occupied with vehicles); mean arrival rate; classified volume for each type of vehicles; incident details like place and duration of incident; and driver behavior probabilities. The model is first tested for homogeneous traffic. The model is then calibrated to represent macrosopic parameters from the field data. A microscopic validation is done using the data generated by VISSIM software. The model provides relations of fundamental parameters, trajectory of individual vehicles, and animation of traffic flow. The output of macroscopic parameters shows that the models are giving satisfactory result.

The second model (TLTF) is an extention of the first to permit lane changing and overtaking. Various lane canging rules were developed by giving due consideration to the drivers behaviour. These model needs additional input of lane changing probability and maximum speed allowed in each lane. The model parameters like driver behaviour probability and lanechanging probability are calibrated. The model is then validated with the field data collected at macroscopic level as well as number of lane changing occurred. The developed model is applied to a long two lane road with the incident occurrence.

The above two models were found to be less effective when there is a wide variation in the vehicular size as each vehicle represented by a single cell. To overcome some of the limitations of these models, the third model (GBTF) is developed which allows the movement of small size vehicles (two-wheelers) side by side. In this model road stretch is divided by grid of cells. Eeach vehicle is represented by one or more cell and the number of cells is governed by size of the vehicle. The vehicle movement rules and data requirement are similar to the previous model. In this model, optimal cell size is decided based on the type of vehicle, clearance, distance headway at various speed, and minimum number of cells by which all kind of vehicle can be represented. The model is calibrated and then validated at microscopic and macroscopic level and found to be more accurate then previous models. However, this model need more computational time as compared to previous model.

The model is used to study the behaviour of heterogeneous traffic like lane changing, incident occurrence, and the effect of different compositon of traffic. This study gives a new computationally efficient model for heterogeneous traffic. However, these models can handle only uni-directional traffic at mid-block. Therefore, further study is required to explore the vehicular behaviour in bi-directional traffic at mid-block as well as intersection.

KEY WORDS: Traffic flow modelling, Simulation, Microscopic,Cellular Automata, Heterogeneous, Homogeneous, Congestion.

A Day Activity Scheduling Procedure for Travel Demand Analysis (2003-06)

Bindhu Muralidhar

with SL Dhingra Completed in 2006

The activity based travel demand analysis has seen substantial development in the past few years. Over these years, various models of activity scheduling behaviour have been formulated and these models are typically part of more comprehensive activity-based models of travel demand. However, studies considering the characteristics of the people in the developing countries are minimal. In this context, the objective of this study is to develope a day activity-scheduling model for activity pattern analysis. To achieve this objective certain set of strategies are formulated. These include the design and administration of activity diaries, the developments of a microsimulation model for activity pattern analysis and scheduling, and the development of a tour mode choice model and a time of day model. The study area selected for the development of the activity-scheduling model is the Thane municipal corporation area of Mumbai metropolitan region, India.

A new type of survey instrument, called time-space diary, is designed combining good features of travel and time-use diary. First, a pilot survey is conducted to assess the suitability of the instrument and various method of survey administration based on the issues like non-response, missing data and cost of survey. Four methods, namely telephonic interview, delivered and mail back method, drop off and pick up method and face-to-face interview method are tried for the administration of the time-space diary. Finally, a main survey is conducted to collect two-day diary information of every individual, above five years, through a household personal interview. The survey was conducted in about 2500 households.

The prototype microsimulated activity-scheduling model developed in this study is for generating activity schedules and travel patterns for a 24-hour typical weekday and a holiday for all the persons in a household. A sequential object oriented approach is adopted for developing synthetic daily individual activity-travel patterns. This approach decomposes the entire daily activity-travel pattern into various components, namely, activity type, activity duration, activity location, time of the day and mode choice. The model makes use of the concept of the tour, a .container. of activities with a common purpose (subsistence, maintenance or leisure), to organize activity episodes into the schedules of persons in a household. The model also takes care of the interaction between the household members. In addition, tour mode choices are predicted based on certain rules of the trip chains within which they occur. The micro-simulated model is validated at the macro level. Validation is also done on the basis of its ability to reproduce the travel parameters.

The output from the microsimulated model is used for the formulation of two models, tour mode choice model and time of day model. The current model is developed only for the subsistence tours. A multinomial logit model is used for the model development. A sensitivity analysis is also carried out for both the models.

One of the innovative and novel approaches of this study is the design of time space diary in the context of a city in a developing country, which combines the advantage of both travel and time use diary format. In addition, microsimulation, an emerging and a promising approach to model travel behaviour is used for the development of activity scheduling model, suitable for a developing country which is minimally explored till now.

Key words: Microsimulation, Activity based model, Scheduling, Tour mode choice model, Tour time-of-day model.

Transportation network optimization (2002-)

B Mallikarjuna Setty

with SL Dhingra

Large scale urbanization and rapid growth of human population as well as vehicle population have laid severe stress on the existing urban transport system in India. It is estimated that by the turn of the century 35% of the total population in India will be living in urban areas. As per 2001 census, the number of Million Plus population cities in India is 27. With the sharing of the limited road capacity by a variety of modes in most cities, the problem has become unmanageable resulting in traffic congestion, inadequate parking area and environmental deterioration. Magnitude of the problem can be reduced by proper planning and control of the available infrastructure facilities.

There is a scope to reduce magnitude of the ill effects of the traffic congestion by simulating and analyzing the congested urban road networks by integrating the techniques used in transportation planning and traffic control. The models developed to integrate the transportation planning and traffic control in traffic assignment phase is popularly called as .Combined Traffic Assignment and Control (CTAC)..

If the traffic congestion can not be mitigated even by demand management using CTAC solutions, there would be a need for capacity enhancements and addition of links to increase the network supply. Capacity enhancements usually carried out under short term improvement programme and addition of links is carried out under long term improvement programme. This is called Network Design Problem (NDP).

Most of the network analysis software like SATURN, EMME/2, TRIPS etc. needs lot of input data preparation and interpretation of the results exogenously for analyzing the transport network with respect demand and supply management measures. Hence, there is a need for development of a single model which can integrate the CTAC (for demand management measures) and NDP (supply management measures) for effective analysis of transport networks.

In this research, refined integrated model of CTAC and NDP problem, application of NDP sub model to Pune city network, field studies carried out for two component link cost functions and two component link cost functions developed for different carriageway configurations, research progress during previous year and further research activities are presented.

Transportation Network Design considering Environmental Parameters and Demand Uncertainty (2004-09)

Sushant Sharma

Transportation network design problem attempts to find the optimal network expansion policies under budget constraints. This can be formulated as a bi-level optimization problem: the upper level determines the optimal link capacity expansion vector and the lower level determines the link flows subject to user equilibrium conditions. The lower level problem represents the driver's behavior in the sense that how he chooses the route. This behavior is represented using the classical Wardropian user equilibrium principles. The objective of this study is to find optimal capacity expansion of large transportation networks by solving it considering deterministic, stochastic and robust network design incorporating both travel time and emissions as performance criteria. Accordingly six variations of the network design problems are formulated and solved for a network spanning from small example network to large scale network.

The first proposed model is a capacity expansion problem of finding optimum capacity vector within the budget constraint and deterministic user equilibrium assignment at lower level. The upper level has been solved using genetic algorithm and lower level by Frank Wolfe algorithm. A comparison has been drawn with various available and attempted heuristics solutions. The results after solving a small and medium sized network shows the efficiency and ability of genetic algorithm to reach a near optimal solution. Sensitivity analysis of solution obtained has been shown with variation in the demand. Finally, the model has been applied in a real network of Pune city of size (370 nodes and 1131 links) to show the efficacy of the model to handle large networks. The major contribution in this part of research work is application of model to a large city network.

The second set of proposed models are extension to first model by including vehicular emissions at lower level and upper level respectively. Recent research findings have shown that travel time variables are affected differently from vehicular emissions, due to traffic flow improvement strategies. So these proposed models are an attempt to reduce emissions while performing network design. The second model is an attempt to solve network design problem when the user is environment cautious. This model modifies the link cost function to incorporate both link travel time as well as emission concerns. Similarly, the third model considers planner as environment cautious and plans the optimal capacity of road network while minimizing health damage cost due to pollutants as well as total system travel time. While developing the following models it was found to realistically quantify emissions there is a need of speed dependent emission factors. Therefore, an experimental study was done to develop speed dependent emission factors. Emission factors in the form of second degree polynomial with speed as the dependent variable were developed for three categories of vehicles and four pollutants namely CO, CO2, NOx, and HC.

Later on, the fourth model named Multi-objective network design for emission and travel time trade off has been proposed. The model uses revised version of fast and elitist non-dominated sorting genetic algorithm as the optimization tool to solve the problem. The model is tested on a small, medium, and large sized test networks. The pareto-optimal solutions generated can act as a trade off between total emissions and total system travel time to account for the planner's desired objectives. The reduction in amount of emission produced while performing capacity expansion proves the necessity of considering emissions along with travel time in network design problem.

The last part of the study i.e. fifth model deals with efficient solution techniques for network design under demand uncertainty. Presently most of such studies needs solving for large number of demand samples thus making them computationally intensive. Therefore, to overcome this limitation various efficient approaches to reduce the sample size of demand for obtaining a demand resilient network design solution have been presented. Specifically, sampling techniques for robust as well as stochastic network design problem and single point approximation and solution construction approach have been used to solve stochastic network design. The results shows the best technique for robust and stochastic network design under demand uncertainty for various demand distributions. The sixth model is a multi-objective formulation for robust network design which is solved using efficient multi-objective genetic algorithm. These studies showed the possibility of reducing the computation time and reaching a quality design solution for network design problem under uncertain demand.

Traffic network design problem attempts to and the optimal network expansion policies under budget constraints. This can be formulated as a bi-level optimization problem: the upper level determines the optimal link capacity expansion vector and the lower level determines the link flows subject to user equilibrium conditions. The lower level problem represents the driver's behavior in the sense that how he chooses the route. This behavior is represented using the classical Wardropian user equilibrium principles.

The first proposed model is a capacity expansion problem of finding optimum capacity vector within the budget constraint and Deterministic User Equilibrium assignment (DUE) at lower level. The upper level has been solved using Gentic Algorithm (GA) and lower level by Frank Wolfe algorithm. A comparision has been drawn with various available and attempted heuristics solutions. The results after solving a small and medium sized network shows the efficiency and ability of GA to reach a near optimal solution. Finally, the model has been applied in a real network of Pune city of size (370 nodes and 1131 links) to show the efficacy of the model to handle large networks. The major contribution in this part of research work is application of model to a large city network.

The second proposed model is extension to first model by including emission pricing at lower level. Traditionally, path travel time is identified as the sole criterion in choosing a particular route. However, in the context of emission pricing, driver's route choice includes travel time as well as environmental concerns. This model is an attempt to solve network design problem when the user is environment cautious. The problem is formulated as a bi-level continuous network design problem with the upper level problem determines the optimal link capacity expansions subject to user travel behavior. The link cost function is modified to incorporate both link travel time as well as emission concerns. Although the link cost function is modified, it is still convex and is therefore solved using convex combination method or Frank-Wolfe algorithm. The upper level problem is an example of system optimum assignment and can be solved using any efficient optimization algorithms. To illustrate the working of the model a small example network has been solved by the proposed algorithms. The reduction in amount of emission produced while performing capacity expansion proves the necessity of emission pricing. Then a medium sized network of Fort Area, Mumbai was further tested to study various computational and sensitivity analysis. The results further give direction to future work to evolve up with speed dependent emission factors. The speed dependent emission factor will give a better view of reduction in emission with improvement in travel time while performing optimal capacity expansion.

In addition to the above models, a new model for reliabilty analysis of transportation network has been proposed by modifying the basic model. This model at upper level has a objective of maximising the network reliabilty of a transportation network rather than optimally expanding the network.

Traffic Flow Modelling Using Cellular Automata for an Urban Road Network (2004-)

R Padmakumar

Researches on traffic flow modelling were concentrated on various areas like mid-block, intersections and ramps, and many of the works were on homogeneous traffic conditions. In developing nations like India, the traffic scenario is even more chaotic, as the element of heterogeneity is brought in by varying vehicular, human, road and environmental factors. Simulation have established as a very vital tool for the traffic analyst, in getting to know more about the traffic flow and to verify the performance of control systems. Cellular automata(CA), a new entrant to this area, introduced a concept of minimal modelling, against the complex modeling procedures being in use till then. It has proven to be computationally efficient in analysing mid-block traffic for homogeneous and heterogeneous conditions.

A few works have been done using CA based simulation on intersections and networks too, but most of them are single-lane, with homogeneous traffic conditions, leaving a wide area open for exploration in the heterogeneous traffic flow in developing countries. The ultimate objective of this study is to develop a computationally efficient network level traffic flow model using CA for urban conditions prevailing in developing nations like India. The model is developed in phases, starting with a four legged signalized intersection as the base module. A network model is then to be built. The model is to be validated with real field data, and compared with established simulation models.

Real-World Driving Cycle for Constructing Emission Factor and Inventory Model to Estimate Vehicular Emissions for an Urban Region (2004-2012)

Sangh Priya (04404805)

Increased number of vehicles on limited road space has resulted in congestion which deteriorates driving condition, thereby further increasing vehicular emissions Various emission control strategies are proposed worldwide with varying degrees of success Design and evaluation of any of these control strategies require reliable emission inventory models to quantify the total emissions from a given region However, the emission modeling in developing country like India is challenging due to heterogeneous traffic characterized by the presence of mixed vehicle type and non-lane based vehicle movement Mixed traffic consists of vehicles ranging from hand driven vehicles such as bicycles, cycle-rickshaws to fuel driven vehicles like motorised two-wheelers and cars Each of these vehicles undergoes diverse acceleration and deceleration characteristics Further, narrow roads, irregular road geometry and absence of lane markings result in chaotic traffic situation

Emission inventory models estimate emissions using vehicular activity data and emission factors The accuracy of the inventory models, therefore, greatly depends on the accuracy of the emission factors and information on vehicle activity The emission factors derived from the real-world driving cycle can provide better estimates of emissions for an urban area Hence, the critical component of all emission models is a driving cycle representing the traffic behaviour Although, standard/ certification Indian driving cycles for motorized two-wheelers and modified Indian driving cycle for cars were developed to test the compliance of Indian vehicles to the relevant emission standards, they neglect higher speed and acceleration and assume all vehicle activities to be similar Therefore, in this study, an attempt has been made to develop a real-world driving cycle using micro-trips extracted from real-world data for estimating emission factors for selected pollutants for an urban region Micro-trip is an excursion between two successive time points at which the vehicle is stopped (idle) This part of motion consists of acceleration, cruise and deceleration modes By convention, a period of rest is at the beginning and end of a micro-trip Each vehicle trip representing speed-time profile of different operating modes of vehicle operation is divided at each idle value.

The uniqueness of this methodology is that the driving cycle is constructed considering five important parameters of the time-space profile, namely, the percentage acceleration, deceleration, idle, cruise, and the average speed and is based on new concept of micro-trips Therefore, this approach is expected to be a better representation of heterogeneous traffic behaviour The real-world driving cycle for the city of Pune in India is constructed for cars and motorized two-wheelers and is compared with existing synthesized cycles Since Pune city has the highest motorised two-wheelers followed by cars and due to limited resources, real-world driving cycles were developed for cars and motorised two-wheelers only However, severe acceleration and deceleration modes of the real-world driving cycle make them difficult to drive on chassis dynamometer thereby limiting its acceptability.

In order to demonstrate the feasibility of testing real-world driving cycle on chassis dynamometer and to develop the emission factors, a chassis dynamometer test has been performed Further, the same vehicles were tested using the conventional driving cycle and the measured values of exhaust emissions was compared Four vehicles, namely, two passenger cars and two motorized two-wheelers each of different age brackets representing old and new technology, was randomly selected from the field and tested on the chassis dynamometer It was observed that the emission factors obtained with the real-world driving cycle are most of the time higher than the synthesized cycles for a given vehicle Therefore, real-world driving cycle gives better insight to the relationship between micro-level driving behaviour and emissions which will be useful in various emission control measures.

The emission factors so obtained from the chassis dynamometer tests and vehicular activity data collected are further used as input parameters in inventory models for estimating total emissions A survey format (questionnaire) was designed to collect vehicle factors and its utilization characteristics that have major influence on emission at selected fuel stations in the vicinity of major intersections The data such as vehicle type, make, model year, fuel used, annual distance traveled, vehicle mileage, average speed, fuel consumption and fuel efficiency was collected from the primary survey Approximately 10,000 data samples were collected from the 21 fuel stations covering different categories of the vehicles such as motorized two-wheelers, cars, taxies, three-wheeled auto-rickshaws, buses, light commercial vehicles and Heavy commercial vehicles The vehicular activity data and emission factors obtained above was used to build two emission inventory models, namely, Average Speed based Emission Model and Fuel Consumption based Emission Model The former involves use of emission factor, cold start, vehicle related parameters like number of vehicles, age of the vehicle, fuel type, average speed and average period of vehicle operation per day The corrections are made in the abstract of the revised report as: The latter uses the fuel consumption and vehicle mileage other than emission factor and number of vehicles These models improve on existing models that are not completely suitable for Indian roads (narrow roads, variation in width of the roads, structural and function condition of roads), traffic conditions (congested, absence of lane discipline) and heterogeneity of vehicles in India This study presents an air quality assessment for a typical urban area like Pune city, India.

Keywords: Driving cycle, Driving characteristics, Micro-trips, Matrices, Chassis dynamometer, Emission factor, emission inventory modeling and Emissions

Analysis of Heterogeneous Traffic on Multilane Highways (2006-)

K.V.R. Ravi Shankar

In a stream of traffic, the behavior of a driver is influenced by the other vehicles and the task of a vehicle following another vehicle under controlled conditions is widely known as car-following behavior. This particular driving task of car-following is of interest because it is relatively simple compared to other driving tasks, has been successfully described by mathematical models. Car following models replicate behaviour of a driver following another vehicle in terms of mathematical relationships and are extensively used in the development of traffic simulation tools. Extensive studies have been conducted in the past five decades to understand car following behaviour resulting in the development of several models. However, all of them are confined to homogeneous traffic and calibrated primarily using test track data. No major study has been reported on car following behaviour in heterogeneous traffic conditions. Heterogeneous traffic comprises of slow and fast moving vehicles and the following behaviour may depend on the type of lead vehicle. Also the vehicular characteristics are different for the typical vehicles like car, bus, and auto. So the present study was initiated with car following behaviour analysis with different combination of vehicles which may find application in modifications of existing microsimulation models.

In the present study, several popular and widely used car following models are evaluated using substantial field data collected under hetergeneous traffic conditions. The data was collected using vehicles equipped with global positioning system (GPS) receivers under heterogeneous traffic conditions from two urban arterials in India. The models are evaluated by comparing several measures of error between the observed and computed model values. The error values are computed after calibrating the model specific parameters by minimizing the error using Genetic algorithm (GA) as the optimization tool. Subsequently, the models are validated with the calibrated parameters using different data set. All these analysis are done separately on data sets corresponding to different lead and following vehicular combinations. Six vehicle combinations are selected by considering car, auto rickshaw, and bus as lead and following vehicles. The desired spacing of vehicle are different for each vehicular combination, indicating the existence of distinct car following behaviour. Similar conclusions are derived from the calibrated sensitivity coefficients for each vehicular combination. From the study, Gipps and Krauss models emerged as the best models and Theil's coefficient as the most suitable measure of error. Based on the results of car following models studied, few modifications to the existing models are suggested. In real traffic conditions, driver may be observing not only the preceding vehicle but a line of vehicles in front. So the study of platoon behaviour consisting of multiple vehicles may be of interest. This is particularly important in heterogeneous traffic conditions where the fast moving vehicles form a queue behind slow moving vehicles leading to traffic jam. Finally the report ends with a lead to the study of this particular platoon stability also called asymptotic stability behaviour in heterogeneous traffic.

Masters Dissertation

Combined traffic assignment and signal control (2004-05)

Mukund Jayant Nashikkar

The conventional traffic assignment models do not consider the junction delay due to signal control. Whereas the signal design assumes constant link flows. Thus the signal setting affects the traffic assignment and visa versa. This interdependency is not considered when the signal setting and traffic assignment are done in isolation. The mutual interaction between user route choices and signal control decisions to arrive at network traffic equilibrium is investigated in this study.

Accordingly, a bi-level optimisation model is proposed. The lower level problem represents Wardrop's user equilibrium model, which incorporated driver's reaction to a given signal control pattern. The upper-level problem is to determine an optimal green split that minimizes the system objective function of total network travel time, taking account of driver's route choice behavior in response to signal split changes. The lower level problem is formulated as a user equilibrium problem and solved using the Frank Wolfe algorithm. The solution to the upper level problem is obtained using genetic algorithm. The link travel time function is modified by adding the signal delay term retaining the convexity property of the function. Two variations of the model are proposed; the first optimising the green split and the second optimising the green split and cycle time.

Both models are tested on a small example network. The study on the example network demonstrated the working of the new combined traffic assignment and signal control problems and the need for combining the present network equilibrium assignment techniques with the signal setting to get a better assignment solution.

The models are also applied for a real medium sized network of the city of Bombay for two peak hours. The results indicate the changes in the link flows and the reduction in the system travel time. The second model performs better in terms of a lower system travel time and smooth convergence of link flows.

Travel time prediction algorithms when using loop detectors

Saurabh Gupta

Evaluation of Area Traffic Control System (2006-)

Suraj H. Shinte

Indian metropolitans have evidenced rapid rate of growth from decades due to surging economy. Traffic demand has increased tremendously due to increase in population as well as vehicle ownerships. Cities are expanding and becoming overcrowded limiting the breadth of road network.

In order to cater the peaking demand it has become essential to adopt effective Traffic Management Systems. One of such aspect is Area Traffic Control (ATC) system. Presently ATC systems like SCOOT and SCAT are popular in advance countries but such systems cannot cope up with Indian situations without adaptation to Indian traffic scenario.

An indigenous Area Traffic Control System is developed by Center for Development of Advance Computing known as Composite Signal Coordination Strategy (CoSiCoSt). Study presented herein aims at evaluating this system and its effectiveness in Indian scenario.

For evaluation data was collected and by using Highway Capacity Manual (HCM) procedure delay is computed. A spreadsheet is modeled which computes critical demand to capacity ratio, degree of saturation of intersection, delay per lane group, delay per approach and delay for intersection as a whole. However the results computed where much on higher side the reason being insufficient data and factors directly adopted from HCM.

Further, heterogeneous traffic at one of the junction in the selected corridor is simulated in VISSIM. This software is based on car following theory. VISSIM provides wide range of results like average travel time, speed, delay, length of queue, delay caused due to transits etc. However, the results obtained cannot be directly relied upon without being compared with actual delay.

In order to get reliable results it becomes essential to crosscheck and calibrate HCM models as well as VISSIM accordingly. For this an exhaustive survey is planned for collection of actual field delay and data to for calibration of HCM and VISSIM. Once the models are calibrated, delay and other performance measure are estimated for coordinated movements along the corridor. Survey for measuring delay after implementation of the CoSiCoSt will be done and impact of Area Traffic Control System will be evaluated.

System Architecture for Area Traffic Control System (2007-08)

Kajal Dubey (06304005)

The objective of this study is to develop an architecture for an area traffic control system suitable for heterogeneous traffic conditions. Scope of the study includes:

  1. Critically evaluate the system architectures of the popular ATCS systems
  2. Propose a system architectures best suited for heterogeneous traffic conditions
  3. Study the major traffic components of the system specifically the vehicle progression models and saturation flow models with the help of field data
  4. Explore the feasibility of a simulator like VISSIM for pre-implementation evaluation of ATC systems

Evaluation of Area Traffic Control System (2007-08)

Revade Ravi Sukhadev (06304020)

The broad objective of this study is to develop methodologies for valuating the performance of ATC systems under heterogeneous traffic conditions. For this study, ATC system deployed by CDAC, India, namely Composite Signal Control Strategy (CoSiCoSt) is selected. To achieve above objective, the scope of study are as follows:

  1. Review of literature focusing on the past studies on the evaluation methodologies, performance measures for the evaluation, and calibration of delay and simulation models
  2. Intersection data collection using video cameras and data extraction
  3. Calibration and validation of HCM delay models and VISSIM simulation models for heterogeneous traffic conditions
  4. Development of methodologies using the calibrated delay and simulation models for evaluating the performance of CoSiCoSt, an ATC system for heterogeneous traffic

Equilibrium Network Traffic Signal Settings (ENETS) (2007-08)

S Nanda Kumar Reddy (06304006)

This study mainly attempts to develop a methodology to integrate the transportation planning model and a traffic engineering model. The individual consideration of the models and the optimization is no longer useful in the field of transportation and especially in congested networks. This motivated to consider the combination of these models to have a flow of traffic in a congested network and decrease the delay at intersections and travel time on links. This is obtained with effective consideration of the offsets concept with signal coordination in the signal control model. A problem is formulated with constraints. The problem is handled by using the optimizing technique .This technique is used to optimize a network performance index, the total network travel time. This total travel time includes link travel time and junction delays. This problem is solved in two stages, referring the both stages as upper stage and lower stage. The upper stage is dealt with the signal control and lower stage is dealt with User Equilibrium assignment. The study is mainly focused on small and medium sized networks.

Public Transit Fleet Management and Information System (2007-08)

Valekar Pritiben Sureshbhai (06304032)

The objective of this study is to develop a public transit fleet management considering the variability in the travel time and to develop a passenger information system suitable for trip planning. The scope of the study includes:

  1. Develop a methodology for fleet scheduling of public transport considering the variability in travel time
  2. Develop a dynamic passenger information system considering the user preferences and suitable for trip planning
  3. Demonstrate these methodologies using selected routes of BEST, Mumbai.

Permanent Traffic Data Collection and Monitoring System (2007-08)

Guru Charandas Nunna (06304801)

The objective of this project is to develop a loop detector based permanent traffic data collection and video surveillance system for an urban arterial. The scope of the study include:

  1. Study the working principle and specification of a loop detector based traffic data collection system
  2. Perform laboratory and field tests with loop detectors to measure traffic parameters with acceptable accuracy, precision, and repeatability
  3. Customizing the Loop detectors for Vehicle count in Indian traffic condition which is heterogeneous and with limited lane discipline
  4. Develop and intelligent tool to supply the historic traffic information for transportation planners
  5. Develop and deploy an intelligent on-line web interface to exhibit the current traffic condition and congestion level on a given travel corridor
  6. Evaluate the accuracy of the vehicle count and classification system using video data.

Effects of Route Choice Models on Network Travel Time Reliabilities (2006-07)

Anurag Kushwaha (05304011)

Travel time reliability is increasingly being recognized as a major factor influencing travel decisions and, consequently, as an important performance measure in transportation management. In this project study a thorough and detailed review of transportation network reliability measures is presented. The study focused on the two most important kinds of network travel time reliability namely the OD travel time reliability and the total system travel time reliability. A new procedure is used to simulate variability associated with the OD travel demand as well as network link capacities and the effect of different route choice models on the value of the travel times (both OD and total system) and on the corresponding travel time reliabilities. Numerical results are presented to examine what the aggregate impact of changes in variability caused by demand and supply variations might have on network assignment. Although the numerical results are based on a small network they clearly show the significance of the route choice model on the estimation of travel time reliability measures. The different assumptions used in modeling route choice behavior could result in different estimates of travel time reliability. Keywords: Travel time reliability, travel time variability, route choice models, traffic assignment.

B. Tech. Project

Transit Route Network Design Using Parallel GA (2001-02)

Jitendra Agrawal

Transit route network design (TRND) problem for urban bus operation involves the determination of a set of transit routes and the associated frequencies that achieves the desired objective. This can be formulated as an optimization problem of minimizing the total system cost: the sum of operating cost and the generalized travel cost. A review of previous approaches to solve this problem reveals the deficiency of conventional optimization techniques and the suitability of genetic algorithm (GA) based models to handle such combinatorial optimization problems. Since GAs are computational intensive optimization techniques, its application to large and complex problems is limited. The computational performance of GA model can be improved by exploiting their inherent parallel nature. Accordingly, two parallel computing models are proposed in this study. The first is global parallel GA model where the fitness evaluation is done concurrently in parallel processing environment. The second is an island parallel GA model where multiple GA population evolve in parallel and exchange the fittest individuals periodically. These models are implemented using both parallel virtual machine (PVM) and message passing interface (MPI), two commonly used libraries. An existing GA model for transit route network design for a large city is used as a case study. These models are tested for speedup and efficiency. From the study it is observed that global model with PVM may prove faster than other models for a similar problem.

Financial evaluation of road projects under bot scheme in developing countries (2002-03)

Saurabh Gupta

The past decade has witnessed many developing countries opening up their economy resulting in greater private sector participation in road infrastructure projects. These countries traditionally followed economic evaluation for the project appraisal. The economic evaluation includes intangible costs and benefits of the project which do not appear in cash flow statement. Hence the economic evaluation cannot indicate the financial performance of a project. Therefore, financial evaluation is important in the context of privatization and commercialization of road projects under BOT scheme. Even though such approaches are very common in developed countries, the financing methods, tax rates and accounting techniques are quite different in developing countries. Therefore, to examine the financial performance of the project and to determine the risks involved, sound financial model are necessary. This paper presents a simulation based financial model for BOT projects. The model takes the length of road, cost of construction volume of traffic, mode wise toll rates and other project specific details as input, performs financial evaluation, and calculates IRR and NPV for project and equity separately. In addition, the model also performs sensitivity analysis and scenario analysis on critical project parameters. A case study of a 40 km road expansion project from Pune (India) has been taken to demonstrate the working of the simulation model. The model gave financially sound project and equity IRR. The sensitivity analysis showed volume of traffic and toll rates as the most sensitive parameters to financial performance for the given project.

Estimating vehicular emission reduction due to network level traffic management (2004-05?)

Harshit Agrawal

With Prof. Virendra Sethi

Urban development and great mobility of people and goods have brought an enormous increase in transportation sector. While on one hand, there has been an undoubted economic growth as a result of these activities; on the other side they have raised severe environmental concerns. Transportation sector is the major source for air pollution in urban areas. Several traffic management schemes are proposed to reduce the impact of air pollution. They may range from micro level management plans like improving road surfaces, road geometry, or better intersection design. At the macro level, they may include schemes like capacity expansion, introduction of alternate mass transportation system etc. An important decision tool in such planning stage is to estimate reasonably accurately the reduction in vehicular emission. This paper presents a generalized methodology to estimate the reduction in emission due to any macro level traffic management schemes. A transport movement takes place by a transport mode on a route. The trip assignment procedure chooses the route to be followed. There are many assignment procedures available and the most popular one is the Wardrop.s User Equilibrium assignment. The basics assumption of this assignment is that a user will select a route that minimizes his/her travel time. The process of assignment selects the routes through the network and loads the trips to the links. The required inputs for assignment are the mode wise origin.destination matrix and network details. The output from the process includes the link flows and system travel costs. Once the link flows are estimated, then the emission can be computed. Based on the trips loaded on the links, the amount of pollutants emitted at the network level is quantified by emission models. Introduction of traffic management scheme will change the travel pattern and will be modeled using the revised origin-destination trip matrix. The above process is repeated and new quantities can be estimated. Emission from the vehicles depends broadly on technical and operational factors of the vehicles, which are quantified in the form of vehicular emission factors. An emission model depends on the travel-related, weather-related, vehicle-related, roadway related, traffic related and driver related factors. The above procedure is illustrated through a case study. The area taken as the case study was the fort area, Mumbai. The origin-destination trip matrix is estimated from the link counts. Link count includes cars, buses and two-wheelers. The OD matrix is assigned to the network and the link flow are computed. From the emission factors of each category of vehicles and the link length the total emission for that link is estimated. This is aggregated to the network level. Then a traffic management scheme is proposed: introduction of a rail corridor. The shift to the rail traffic is estimated by model-split model, thus providing a new OD matrix. This new OD matrix is assigned to the network to compute the new emission quantities for the network. Two pollutants namely CO and TSP (total suspended particulates) is considered in the case study.

Research Topics

Project topics for M.Tech/Ph.D

Second generation Area Traffic Control System (CoSiCoSt2G)

The objective of the project to upgrade the composite signal control strategy (CoSiCoSt2G), an area traffic control system developed by the CDAC (Trivandrum). The scope of the proposed up gradation include development of a real-time link travel time estimator, development of an ATC simulator to model and evaluate CoSiCoSt System, and development of a real time system which supports network level traffic management. The city of Pune is proposed as the implementation site where an ATC system exists.

Real-time Traffic Counting and Monitoring System

The goal of the project is to develop a real time road traffic counting, classifying, and analyzing system along with monitoring of the traffic. The objectives of the project are customizing the Loop detectors for Vehicle count in Indian traffic condition, exploring the efficacy of video-image processing to detect, count, classify, and determine the speed of the vehicles, and to evelop an intelligent tool to analyze the traffic.

Real-time Transit Trip Planner and Route Information System

The project is to develop a real time and intelligent public transit bus trip information system and route information. The specific objectives are the development of a static trip planner, GPS based real-time trip planner, web interface for the users, and route information system. The project is to be demonstrated in transit systems of Pune or Mumbai.

Speed-Flow-Density (SFD) Relationship for Heterogeneous Traffic Conditions

While there are newer models for SFD relationships, due the simplicity of Greenshield's model it is still widely used for portraying SFD and capacity constraints. This project will first conduct a detailed literature review on SFD models and their utility. Then take available data for Indian Cities at midblocks of arterial roads and develop SFD relationships that are a factor of traffic composition. Bachelors Project - two Students, or Masters Project

Timing Signals in Heterogeneous Traffic

Webster's green split calculation methodology applies sat ratio (volume/saturation flow) proration to determine splits for each movement. In India the same methodology is applied using the concept of Passenger Car Equivalents (PCE). Hussain (x) determined that the concept of passenger car equivalents do not exist for heterogeneous traffic. Hussain concluded that capacities are signals are a function of the geometry upstream of downstream of signal in heterogeneous traffic. Hussain developed a simulation based model to determine the capacities at signals. This study will review Hussain's work, and develop a new methodology for timing signals that is not based on PCE and subsequent saturation flow estimation. The study will look at possibility of using density as a proxy for traffic flow and signal timing. Turn-Bay Lengths Masters Project Good Signal Timings do not exist in vacuum; they are a factor of the (1) Location, (2) Intersection Geometry, and (3) User Characteristics. From an engineering point-of-view, of these three the only thing the operator can really control is the intersection geometry. Good geometric design will immensely help timing junctions. Since at most arterial junctions the highest percentage of traffic is through traffic; cycle length and consequently junction delays can be drastically reduced by providing a combined green (Lead-Lead or Lag-Lag phasing) for the opposing through movements. In India, most junctions do not have separate turn-bays for right-turning vehicles. Hence, creating lead-lead or lag-lag phasing causes safety problems. To improve junction delays it is imperative that turn-bays are provided to separate turning vehicles from through vehicles. This project will create guidance for the length of turn-bays based on the number of vehicles and traffic composition. Highway Capacity Manual gives a thumb rule that the length of turn-bay should be equal to number of turning vehicles per hour. Similar rule, with associated volume curves will be the output of this project. The student will use microsimulation analysis as a means to developing these relationships. Ph.D. Project

Trip Generation Manual

The approach used by firms to determine trip generation of a new shopping complex is to do a survey of an existing shopping complex. Multiple studies of existing land use have been done all over the country by various consultancy firms. Unfortunately, none of these studies are documented in the public domain. This project will look at compiling available data from various consultancy firms across the country to develop a Trip Generation Manual for India, on the lines of the ITE Trip Generation Manual. The student will develop relationships (curves) for various land uses, and also conduct additional studies to bridge obvious gaps in the trip generation manual. Master's or Ph.D. Project

Determination of U-Turn Radius of Curvature for U-Turning Vehicles

Medians are commonly used in Indian cities to discipline vehicles from encroaching the opposite direction lanes, and to discourage pedestrians from crossing the carriageway at all places. One of the impacts of the medians is that vehicles that have to turn at midblock junctions that are obstructed by the median have to travel to the next undivided junction and take a u-turn. Since u-turning vehicles stick to the rightmost lane, there is often not enough curvature for the u-turning vehicles to turn without making multiple maneuvers. Knowledge of turning radius of various types of Indian vehicles will allow geometric designers to accommodate u-turning vehicles at the junctions. This study will study seven classes of vehicles (compact car, mid-size car, large car, mini-van, ambulance, bus, fire-trucks) for their u-turn radius of curvature. Master's or Ph.D. Project

Vehicle Actuation Capabilities of Signal Controllers

Loops detectors are commonly used across the world for detecting vehicle presence and passage. Vehicle actuated signals improve compliance to signal indications and give great benefits during non-peak hours and when cross street volumes are low. For vehicle actuated signals to work properly they should be able to detect autos, motor cycles, scooters and mopeds; in addition to the larger vehicles. This study will evaluate the accuracy and efficiency of detection of the detectors available in India. The objective of the study is to develop performance criteria for detectors for usage in the field. Bachelor's Project - two students, or Masters Project

Area Traffic Control Evaluation

Various Indian cities are opting for Area Traffic Control (ATC) to manage their signals. Area Traffic Control works on the premise of creating green waves, that is - vehicles platoons are identified upstream of signals and then allowed to traverse through the signals. This project will document the various ATC systems working in India (SCOOT, CDAC, SCATS, etc.), and evaluate the performance of the ATC systems that are implemented. Masters Project

Developing Behavioral Models for Microsimulation Analysis

While multiple heterogeneous simulation models have been developed over the years in India, none of them account the human behavioral interactions that impact the traffic operations. This project will develop heterogeneous behavioral models for Indian driving culture through trajectory data analysis. Ph.D. Project

Sampling techniques for robust design of infrastructure facilities

Analysis and design of several infra structural system needs to consider the stochastic effect of design parameters. Often, a distribution of how these parameters is assumed and random samples are drawn. Analysis and design is made by evaluating the performance of the system of each random sample and computes various moments like mean, variance, covariance etc., to be used in the design. The problem becomes extremely complex when several stochastic parameters are involved. Sampling techniques randomly draws samples from the distribution in such a way that the moments are with a reasonable error. For instance, a preliminary study of using antithetic sampling has reduced the computation time by 50 times. The project is proposed to developed efficient sampling techniques like antithetic sampling, quasi-Monte-Carlo sampling, and Latin hyper cube to be used in infrastructure design. The study will take two or three infrastruce problems (for instance design a transport infrastructure considering the stochastic nature of travel demand) and propose the more efficient sampling techniques.

Discrete simulation models using cellular automata

The project envisages replacing the time consuming micros simulation method using the power features of the cellular automata, a discrete simulation model frame work. Micro simulation tools typically describe the entire system dynamics using accurate mathematical models whose operations involve precise numerical implementations resulting in huge computation time for any real applications. However, several infrastructure designs do not require such accuracy; and the essential characteristics of a system can be model with simple discreet and dynamic model of cellular automata. The project proposed to develop a network level traffic simulation model and compare the performance with classical micro simulation models.

An intelligent agent based modeling system

This project will assess the ability of an intelligent agent system in the design of infrastructure system. While the proposed research will draw on the modeling and simulation, the prime focus will be to contribute to the research in soft computing techniques. The objective of this research is to test the hypothesis that a team of intelligent agents, encompassing a multi-algorithmic approach to optimization, will more effectively and efficiently solve a typical infrastructure system design problem. This research will explore synergies that exist between different solution techniques and exploit these synergies to improve the system.s performance. This research will also assess the ability of the system to be structured to manage the competing objectives in this multi-criteria optimization problem. The specific problem that will be demonstrating the efficiency and scalability is the design of a large and complex transit information system considering diverse characteristics of travelers and transit units moving over a space and time along some fixed routes.

Understanding the Dynamics of Crowd Behavior in Mumbai Subway and CST Terminal

Abstract: The density of subway travelers in Mumbai is one of the highest in the world. This level of congestion brings about significant challenges in modeling the dynamics of crowd behavior especially at while subway riders want to enter/exit the train. This study will develop fluid based particle models to understand the dynamics of crowd behavior in congested train compartments using data collected from multiple video cameras. The data will be collected during peak periods and non-peak periods both in the general and ladies compartments. The video feed will be analyzed using state of the art image processing techniques to obtain spatial and temporal characteristics of individual travelers. Similar data will be collected at the CST terminal to understand the dynamics of crowd movement in open areas. The insights obtained from this problem will be an important input into agent based simulation models and in the better design of egress for trains and design features in open terminals. (Along with Satish Ukkusuri, RPI, USA).

Some are formulted by Dr. Vijaya, Dr. Ukkusuri,

Seminar topics for B.Tech/M.Tech

Review of hydrodynamic models of traffic flow

Traffic flow is often considered akin to the flow of fluids and varius hydro dynamic models are proposed. This reveiw is intented to highlight various models starting from first order models (LWR) models to the higher order models. Referecences: Kuhne & Michalopoulos (1999)Zhang (1998)Zhang (1999)

Review of car-following models

The car-following models examines the manner in which individual vehicles (and their drivers) follow one another. In general, they are developed from a stimulus-response relationship, where the response of successive drivers in the traffic stream is to accelerate or decelerate in proportion to the magnitude of the stimulus at time t after a time lag T. Car-following models form a bridge between the microscopic behavior of individual vehicles and the macroscopic characteristics of a single-lane traffic stream with its corresponding flow and stability properties. References: Brackstone & McDonald (1999), Aycin & Benekohal (2001) A review of car following models

Review of optimal velocity models

In the Optimal Velocity Model proposed as a new version of Car Following Model, it has been found that a congested flow is generated spontaneously from a homogeneous flow for a certain range of the traffic density. A well-established congested flow obtained in a numerical simulation shows a remarkable repetitive property such that the velocity of a vehicle evolves exactly in the same way as that of its preceding one except a time delay T. This leads to a global pattern formation in time development of vehicles' motion, and gives rise to a closed trajectory on . x-v (headway-velocity) plane connecting congested and free flow points. References: Nakanishi et al. (1996),OVM

Capcity of basic freeway sections under mixed traffic conditions

Capacity analysis of various types of streets and highways which include the ability to analyze expected operating conditions and performance under a range of traffic demands. A number of research efforts are currently underway which will lead better understanding of procedures for estimating cacity of various traffic facilities. For instance a free way elemets include basic sections, weaving areas. and ramps. References: Roess et al. (1980), Highway Capacity Manual (2000).

Urban traffic control by traffic signals

Urban traffic control (UTC) by traffic signals is a major element of safety for vehicles and pedestrians when crossing an intersection. Traffic signals share the time between the different flows, that are considered to be antagonistic, and eliminate the most serious conflicts. Nevertheless, in the face of the ever-growing amount of traffic in urban areas, traffic control by signals has become a significant traffic management tool in order to reduce the consequences of this increase: congestion, delay, stop, pollution, fuel consumption, noise, stress, discomfort. From a methodological point of view, the first major problem when controlling signalized intersections is to face different conflictual objectives This review tries to address the methodological aspects of traffic control and the performce evaluation. References: Boillot et al. (2006).

Comparison of signalized intersection delay models

Vehicle delay is perhaps the most important parameter used by transportation professionals to evaluate the performance of signalized intersections. This importance of vehicle delay is reflected in the use of this parameter in both design and evaluation practices. For example, delay minimization is frequently used as a primary optimization criterion when determining the operating parameters of traffic signals at isolated and coordinated intersections. While many methods are currently available to estimate delays at intersection approaches, very little study has been conducted to assess the consistency of these estimates. This study addresses this issue by comparing the delays estimated by a number of existing delay models for a signalized intersection approach controlled in fixed-time and operated in a range of conditions extending from under-saturated to highly saturated. References: Boillot et al. (2006),Dion et al. (2004).

Network level traffic control

First level of urban traffic control is isolated intersections, and the next level is coordinated singnal at arterials. The highest form of intersection signal control is real time control of all intersections in the network. The input to the system will be the traffic counts by detectors. It is important to uderstand the control algorithms and performace evaluation parameters. References: Mirchandani & Head (2001).

Inductive loop detectors for vehicle detection

Inductive loop detectors can be used to detect the presence of vehicle. It can also be used for speed detection, vehicle re-identification, density estimation, indentifying congestion, travel time estimation etc. References: Coifman & Cassidy (2002), Coifman (2003), (), Coifman (2002), Coifman (2001).

Video images for vehicle detections

Video images are increasingly being used in vehicle detection and supply the information to the traffic controlers. Coifman et al. (1998).

Combined traffic assignment and signal control

Traffic assignment assumes there is only link dealys and ignores any junction dealy. However, the reality is that there are signal delays which invalidates the above assumption. On the contarary, signal design assumes constant flow and ignores the variation of traffic with respect to the signal delays.The recent approach is to combine the classical transportation planning stream (traffic assignment) and traffic engineering (signal setting) stream into one unit. This is expected to be more realistic. Refer: Meneguzzer (1997)Yang & Yagar (1995)Smith (1993).

Modeling vehicular emissions

There has been a substantial growth of road traffic over the years and consequently, of air pollution caused by the vehicular exhausts, which is now been considered as one of the primary source of urban air pollution. Many studies have been conducted in the past to model the vehicular exhaust emission. Considering the magnitude of problem and, the importance of these studies, it becomes quite relevant to have an assessment/review of the existing modelling studies carried out in this area. References: Sharma & Khare (2001).

Air quality modeling at a signalized intersection

Traffic signal ment to control confliting traffic adopts a time sharing strategy. At signals, vehcles had to slow down and may have to stop. This causes vehciles to deccelerate, accelerate, and idle causing enhanced emission. An objective understanding in necessay to evaluate various control strategies. References: Midenet et al. (2004)Sharma & Khare (2001).

Development of driving cycle for Indian conditions

Development of speed dependent emission factors for Indian conditions

An artificial neural network (ANN) is an interconnected group of artificial neurons that uses a mathematical model or computational model for information processing based on a connectionist approach to computation. In most cases an ANN is an adaptive system that changes its structure based on external or internal information that flows through the network. It basically tries to emulate the working of human behavior. This topic tries to explore the application of ANN in transportation. References: Mark (1995)Ishak & Al-Deek (1998)Nakatsuji & Kaku (1991).

Application of genetic algorithms in transportation

Application of cellular automata in transportation networks

How cellular automata models (1-Dimentional models for mid block sections and 2-D models for intersections) can be extented to handle large transporation networks. Chopard et al. (1995), Chopard et al. (1996), Chopard et al. (1999). A brief write of the same

Distributed comuting

Transportation systems has many large network related problems to solve which require large computing time. Although large computing power can be used, a cheaper and viable alternative is to use the distributed computing power. (References: Wong (1997)Lin & Yao (1999)Balla & Lingireddy (2000))

GA-SA hybrid systems

Genetic algorithm is one of the highly popular optimization tool especially for non-convex and ill structured optimization propblem. However, it is limited by large computing time and inability to converge to global optimization problem. On the contaray, simulated annealing, another AI based optimization tool has a good ability to converge to optimal solution. A hybrid GA-SA sytem shows considerable promise and this study is expected to explore this.


Aycin, M. F. & Benekohal, R. F. ( 2001),
``Stability and performance of car following models in congested traffic'', Journal of Transportation Engineering, ASCE , Vol. 127, pp. 1-12.

Balla, M. C. & Lingireddy, S. ( 2000),
``Distributed genetic algorithm model on network of personal computers.'', Journal of Computing in Civil Engineering , Vol. 14(3), pp. 199-205.

Boillot, F., Midenet, S. & Pierrele, J.-C. ( 2006),
``The real-time urban traffic control system cronos: Algorithm and experiments'', Transportation Research Part C , Vol. 14, pp. 18-38.

Brackstone, M. & McDonald, M. ( 1999),
``M brackstone and m mcdonald'', Transportation Research Part F , pp. 181-196.

Chopard, B., Dupuis, A. & Luthi, P. ( 1999),
``A cellular automata model for urban traffic and its application to the city of geneva'', University of Geneva, CUI , 24 rue General-Dufour, CH-1211 Geneva 4, Switzerland.

Chopard, B., Luthi, P. O. & Queloz, P.-A. ( 1996),
``Cellular automata model of car traffic in two-dimensional street networks'', J.Phys. A , Vol. 29, pp. 2325-2336.

Chopard, B., Queloz, P.-A. & Luthi, P. O. ( 1995),
`Traffic models of a 2d road network'.

Coifman, B. (2001),
``Improved velocity estimation using single loop detectors'', Transportation Research Part A: Policy and Practice , Vol. 35, pp. 863-880.

Coifman, B. (2002),
``Estimating travel times and vehicle trajectories on freeways using dual loop detectors'', Transportation Research Part A: Policy and Practice , Vol. 36, pp. 351-364.

Coifman, B. (2003),
``Estimating density and lane inflow on a freeway segment'', Transportation Research Part A: Policy and Practice , Vol. 37, pp. 689-701.

Coifman, B., Beymer, D., McLauchlan, P. & Malik, J. (1998),
``A real-time computer vision system for vehicle tracking and traffic surveillance'', Transportation Research Part C: Emerging Technologies , Vol. 6, pp. 271-288.

Coifman, B. & Cassidy, M. (2002),
``Vehicle reidentification and travel time measurement on congested freeways'', Transportation Research Part A: Policy and Practice , Vol. 36, pp. 899-917.

Dion, F., Rakha, H. & Kang, Y.-S. ( 2004),
``Comparison of delay estimates at under-saturated and over-saturated pre-timed signalized intersections'', Transportation Research Part B , Vol. 38, pp. i99-122.

Highway Capacity Manual (2000),
Transportation Research Board, National Research Council, Washington, D.C.

Ishak, S. & Al-Deek, H. (1998),
``Performance of automatic ann-based incident detection on freeways'', Journal of Transportation Engineering, ASCE , Vol. 125(4), pp. 281-290.

Kuhne, R. & Michalopoulos, P. ( 1999),
Continuum flow models, in H. Lieu, ed., `Traffic flow theory: A state-of-the-art report', FWHA, chapter 5.

Lin, G. & Yao, S. (1999),
Parallel genetic algorithm on pvm, in `Proceedings of the International Conference on Parallel Algorithms', Wuhan, P.R. China.

Mark, D. (1995),
``A review of neural networks applied to transport'', Transportation Research Part C , Vol. 3(4), pp. 247-260.

Meneguzzer, C. (1997),
``Review of models combining traffic assignment and signal control'', Journal of Transportation Engineering , Vol. 123(2), pp. 148-155.

Midenet, S., Boillot, F. & Pierrelee, J.-C. ( 2004),
``Signalized intersection with real-time adaptive control: on-field assessment of co2 and pollutant emission reduction'', Transportation Research Part D: Transport and Environment , Vol. 9, pp. 29-47.

Mirchandani, P. & Head, L. (2001),
``A real-time traffic signal control system: architecture, algorithms, and analysis'', Transportation Research Part C: Emerging Technologies , Vol. 9, pp. 415-432.

Nakanishi, K., Itoh, K., Igarashi, Y. & Bando, M. ( 1996),
`Solvable optimal velocity models and asymptotic trajectory'.

Nakatsuji, T. & Kaku, T. (1991),
``Development of a self-organizing traffic control system using neural network models'', Transportation Research Record , Vol. 1324, pp. 137-145.

Pattnaik, S. B., Mohan, S. & Tom, V. M. ( 1998),
``Urban bus transit route network design using genetic algorithm'', Journal of Transportation Engineering, ASCE , Vol. 124(4), pp. 368-375.

Roess, R. P., Linzer, E. M., McShane, W. R. & Pignataro, L. J. (1980),
``A revised procedure for the capacity analysis of basic freeway sections'', Transportation Research Part B , Vol. 14, pp. 1-11.

Sharma, P. & Khare, M. (2001),
``Modelling of vehicular exhausts - a review'', Transportation Research Part D: Transport and Environment , Vol. 6, pp. 179-198.

Smith, M. J. (1993),
``Traffic equilibrium with responsive traffic control'', Transportation Science , Vol. 27(2), pp. 118-132.

Tom, V. M. & Mohan, S. (2003),
``Transit route network design using frequency coded genetic algorithm'', Journal of Transportation Engineering, ASCE , Vol. 129(2), pp. 186-195.

Wong, S. C. (1997),
``Group-based optimisation of signal timings using parallel computing'', Transportation Research Part C , Vol. 5(2), pp. 123-139.

Yang, H. & Yagar, S. (1995),
``Traffic assignment and signal control in saturated road networks'', Transportation Research Part B , Vol. 29, pp. 125-139.

Zhang, H. M. (1998),
``A theory of nonequillibrium traffic flow'', Transportation Research Part B , Vol. 32B, pp. 485-498.

Zhang, H. M. (1999),
``Analysis of the stability and wave properties of a new continuum traffic theory'', Transportation Research Part B , Vol. 33B, pp. 399-415.

Prof. Tom V. Mathew 2012-08-02