Modeling Managed Lanes in Long-term Planning Models

Sept'17 - Present

Funded by National Central Texas Council of Governments (NCTCOG) university partnership program
  • Identifying ways to model dynamics of toll and route choice on managed lane networks in a static traffic assignment model
  • Proposing recommendations on the ranges of value of time to model different user classes
  • Developing a user equlibrium model with recourse for modeling the online traveler choices

Using National Performance Management Research Data Set (NPMRDS) for Decision Making on US281N corridor

May'17 - Aug'17

Funded by Texas department of transportation, San Antonio district
  • Created a PostgreSQL database of the dataset for the city of San Antonio and ran SQL queries to extract relevant results
  • Developed a R-shiny user interface for the city which lets user extract meaningful information from the dataset. Available online at NPMRDS San Antonio
  • Compared the performance of instantaneous and time-dependent aggregation of corridor travel time
  • Performed k-means clustering using a proposed distance metric to determine the number of atypical days on the corridor

Data Rodeo: the City of Austin

May'17 - Aug'17

Funded by City of Austin
  • Objective: Develop methods to use and integrate different datasets for traffic management and transportation planning in practice
  • Video data
    • Assisted the team at Texas advanced computing center (TACC) develop and train deep learning models for object detection in large scale traffic videos
    • Identified and helped implement use cases for the algorithms: vehicle counting and bicycle-pedestrian safety analysis
  • Data from Bluetooth and WiFi sensors
    • Wrote a Python code which determines travel time on a corridor by identifying a tracked vehicle ID across different WiFi sensors installed
    • Compared the performance of the two datasets against the NPMRDS travel time for the same corridor

Operational Analysis of Active Traffic Management(ATM) Strategies

Jan'15 - Apr'17

Funded by Texas department of transportation project number 0-6859
  • Evaluated the operational impacts of strategies on a network level using VISTA software for Dynamic Traffic Assignment
  • Integrated the results with the microsimulation model of the corridor in VISSIM for ATM Strategies including ramp metering, variable speed limits etc.
  • Developed a spreadsheet tool incorporating regression models which predict percent improvement in corridor travel time given an ATM strategy and the inputs of demand and bottleneck speed

Decentralized modeling of large scale transportation networks for consistent planning models

Aug'14 - Current

Funded by National Science Foundation grant number 1254921 and Texas department of transportation project number 0-6900
  • Programmed and tested an improved decentralized algorithm to perform traffic assignment on large scale networks (called a decomposition approach to static traffic assignment or DSTAP)
  • Developed and tested algorithms that determine network partitions for solving DSTAP more efficiently
  • Provided a set of recommendations for improving consistency between the existing Texas statewide model and the models operated by different Metropolitan planning organization (MPOs)

Developing Simulation model for Indian Road networks using TRANSIMS

May'12 - May'14

Guide - Prof. Gopal R. Patil, Department of Civil Engineering, IIT Bombay
Achievements
  • Presented the paper in International Conference on Agent based modelling in Transportation Planning and Operations held in Virginia Tech Oct'13
  • Presented a poster in the 10th International Conference on Transportation Planning and Implementation Methodologies for Developing Countries (TPMDC'12) held in IIT-Bombay Dec'12
Research
  • Developed a simulation model of the IIT Bombay campus in TRANSIMS with calibrated traffic flow on links accurate to 10% confidence limits
  • Introduced new vehicle classes in the open source code of TRANSIMS written in C++ to get India specific simulation
  • Interviewed and compiled campus household travel survey and assisted in preparing an activity based model for the campus
  • Undergraduate Thesis: Developed and adapted TRANSIMS parameters for simulation results to match driving practices prevalent in India like left side driving and varying proportion of different vehicle classes

Traffic Light metamodel for MATSIM Singapore

May'13 - July'13

Guide - Prof. K. Axhausen & Prof A. Erath (Eth Zurich - Future Cities Laboratory, Singapore)
  • Developed a model for incorporating signal delays in MATSIM, an agent based traffic simulation model leading to reduced computation time and simplification of inputs for signal detailing
  • Programmed around 1000 lines of code using MATSIM listeners and controllers, handling agent events in Java platform
  • Compared the existing input-intensive method to the proposed model results and identified shortcomings of the proposed approach over different others
Web and coding club (WNCC) IIT Bombay | Self mentored
  • Developed an innovative three-dimensional game in C++ with OpenGL using GLUT managing 1500 lines of code and testing the game successfully for different test conditions
  • Presented the work in annual technical exhibition conducted in Sept'11, IIT Bombay
  • The wiki page documentation is available here

Traffic Intersection modelling in AnyLogic

Guide- Prof. J.Venkateswaran, IEOR Department, IIT-Bombay
  • Built and structured a model of traffic intersection in ANY-LOGIC to judge effectiveness of different signal schemes for a real demand data collected at an intersection in Pune city
  • Optimized the signal timings for minimizing delay using Opt-quest tool and reported the optimal signal plan