Presentation-Kachroo Theory Based Traffic Safety.pdf

Viewer
Transcript

Feedback Theory based Traffic Safety Campaigns

Feedback Theory based Traffic Safety Campaigns Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Masha Wilson (Ph.D. OTS, Nevada

January 24, 2011 Office of Traffic Safety, Nevada Nevada University Transportation Center

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Outline

1 Introduction

Designing feedback based Campaigns 2 Data Collection Process 3 Data Sensors 4 Analysis 5 Data Visualization 6 Feedback Control Design

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Introduction Designing feedback based Campaigns

Feedback Control

Target Behavior

Difference between the two Media Campaign Behaviors and Education

Media Campaign

Design (Controller)

Drivers and their Behavior (System)

Actual Behavior

Measured Behavior Sensor

Seat Belt data, crash data, traffic data etc.

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Introduction Designing feedback based Campaigns

Designing feedback based Campaigns for Daytime Seat-belt Usage

Event: Click It or Ticket (CIOT) Mobilization during May, focusing upon increasing seat-belt usage.

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Introduction Designing feedback based Campaigns

Designing feedback based Campaigns for Daytime Seat-belt Usage

Event: Click It or Ticket (CIOT) Mobilization during May, focusing upon increasing seat-belt usage. Features: Increased law enforcement, broadcasting messages on television and radio.

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Introduction Designing feedback based Campaigns

Designing feedback based Campaigns for Daytime Seat-belt Usage

Event: Click It or Ticket (CIOT) Mobilization during May, focusing upon increasing seat-belt usage. Features: Increased law enforcement, broadcasting messages on television and radio. Example: North Las Vegas Police will be running a ’Click It or Ticket’ Campaign today from 11 am

to 9 pm and be warned, there will be a zero tolerance policy in place (KTNV-ABC,LAS VEGAS, NV JUN 27 2010, 6:00PM, Television Monitoring Services)

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Introduction Designing feedback based Campaigns

Designing feedback based Campaigns for Daytime Seat-belt Usage

Event: Click It or Ticket (CIOT) Mobilization during May, focusing upon increasing seat-belt usage. Features: Increased law enforcement, broadcasting messages on television and radio. Example: North Las Vegas Police will be running a ’Click It or Ticket’ Campaign today from 11 am

to 9 pm and be warned, there will be a zero tolerance policy in place (KTNV-ABC,LAS VEGAS, NV JUN 27 2010, 6:00PM, Television Monitoring Services)

Feedback based campaign design required, to increase efficiency.

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Data Collection Process

Data Collection Variables

Front Seat Occupants: Drivers and Passengers.

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Data Collection Process

Data Collection Variables

Front Seat Occupants: Drivers and Passengers. Age: (<15, 15-19, 20-60, >60)

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Data Collection Process

Data Collection Variables

Front Seat Occupants: Drivers and Passengers. Age: (<15, 15-19, 20-60, >60) Gender: Male, Female

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Data Collection Process

Data Collection Variables

Front Seat Occupants: Drivers and Passengers. Age: (<15, 15-19, 20-60, >60) Gender: Male, Female Ethnicity: Caucasian, Hispanic, African-American, Others

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Data Collection Process

Data Collection Variables

Front Seat Occupants: Drivers and Passengers. Age: (<15, 15-19, 20-60, >60) Gender: Male, Female Ethnicity: Caucasian, Hispanic, African-American, Others Vehicle Type: Sedan/Station Wagon, Van/SUV, Pickup.

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Data Collection Process

Data Collection Variables

Front Seat Occupants: Drivers and Passengers. Age: (<15, 15-19, 20-60, >60) Gender: Male, Female Ethnicity: Caucasian, Hispanic, African-American, Others Vehicle Type: Sedan/Station Wagon, Van/SUV, Pickup. State of Registration: Nevada, California, Others.

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Data Sensors

Data Sensors - Unconstrained Space Problem: How fast can you collect the sea-belt data for front seat occupants at a site where cars pass by at 70 m.p.h ?

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Data Sensors

Data Sensors - Unconstrained Space Problem: How fast can you collect the sea-belt data for front seat occupants at a site where cars pass by at 70 m.p.h ? For an unconstrained design space, just one user input is sufficient.

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Data Sensors

Data Sensors - Unconstrained Space Problem: How fast can you collect the sea-belt data for front seat occupants at a site where cars pass by at 70 m.p.h ? For an unconstrained design space, just one user input is sufficient. Examples of designs on unconstained space: X

X Y

5

Z4

Z2

Z3

X

X

Y

Z

Y

Y

Z1

Y X Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Data Sensors

Data Sensors: Constrained Space of PDA/Smartphones Development on PDA and Smartphone - data collection in 4 clicks (worst case) when compared with 11 clicks (Vivoda and David, 2006) and 17 clicks (Nevada, 2006-2009).

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Data Sensors

Data Sensors: Constrained Space of PDA/Smartphones Development on PDA and Smartphone - data collection in 4 clicks (worst case) when compared with 11 clicks (Vivoda and David, 2006) and 17 clicks (Nevada, 2006-2009). (a) PDA - C# and Ultralite - Free. (b) IPhone - Objective C and Sqlite on Xcode - Free.

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Data Sensors

Data Sensors: Constrained Space of PDA/Smartphones Development on PDA and Smartphone - data collection in 4 clicks (worst case) when compared with 11 clicks (Vivoda and David, 2006) and 17 clicks (Nevada, 2006-2009). (a) PDA - C# and Ultralite - Free. (b) IPhone - Objective C and Sqlite on Xcode - Free.

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Data Sensors

Adaptively minimizing user input Static default selections in the previous template can be replaced by adaptive selections.

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Data Sensors

Adaptively minimizing user input Static default selections in the previous template can be replaced by adaptive selections. Based on the neighbourhood, the default selection will change over time.

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Data Sensors

Adaptively minimizing user input Static default selections in the previous template can be replaced by adaptive selections. Based on the neighbourhood, the default selection will change over time. Techniques being implemented: 1

2

Exponential/Discrete forgetting based competitive countingGiven M to be a fixed look back in time, xi being avariable mutually exclusive than other N − 1 variables, ∀xi ǫ 0, 1 , P0 select ith variable with maximum value of k=−M αe βk xi [k] in P0 continuous sense or k=−M αaβk xi [k] in discrete sense. Bayesian learning based probabilistic models of variables.

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Analysis

Statistical Analysis on Collected Data Finding the weighted statewide seat-belt usage using bootstrap sampling.

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Analysis

Statistical Analysis on Collected Data Finding the weighted statewide seat-belt usage using bootstrap sampling. Open Source R: Bootstrap sampling, and histogram plotting of sample means of difference in Post-Mobilization and Pre-Mobilization seat-belt usage.

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Data Visualization

Data Visualization: Geolocation of sites on Google Maps

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Data Visualization

Data Visualization: Site-Specific Analysis

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Data Visualization

Data Visualization: Interactive Animated Plots for every site

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Data Visualization

Data Visualization: Building Statewide Custom Plots

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Data Visualization

Data Visualization: Example of Age and Gender based Statewide Seat-belt Usage

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Feedback Control Design

Feedback Control

Target Behavior

Difference between the two Media Campaign Behaviors and Education

Media Campaign

Design (Controller)

Drivers and their Behavior (System)

Actual Behavior

Measured Behavior Sensor

Seat Belt data, crash data, traffic data etc.

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Feedback Control Design

Where Nevadans read, saw or heard about Seat Belt Law Enforcement (n = 341)

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Feedback Control Design

Qualitative, Quantitative and mixed analysis on TV data TV data:

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Feedback Control Design

Qualitative, Quantitative and mixed analysis on TV data TV data:

Extensive Comparison of softwares: http://www.restore.ac.uk/lboro/research/software/caqdas_comparison.php Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Feedback Control Design

Example of Coverage areas of a TV station (KOLO-ABC, RENO)

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Feedback Control Design

Epidemology based Diffusion Modelling

∂S = −βSI + a(S + R) ∂t ∂2I ∂I ∂S =D 2 −v + βSI − dI − rI ∂t ∂x ∂x ∂R = rI ∂t S(x, 0) = S0 I (x, 0) ∝ δ(x) R(x, 0) = 0

(1)

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Feedback Control Design

Example of 2-D Diffusion

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Feedback Control Design

Transportation Networks: Connecting People

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Feedback Control Design

Conclusion

Feedback control framework.

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Feedback Control Design

Conclusion

Feedback control framework. Component design: Sensor, campaign (Online and offline visualization, animations and analysis)

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns

Feedback Theory based Traffic Safety Campaigns Feedback Control Design

Conclusion

Feedback control framework. Component design: Sensor, campaign (Online and offline visualization, animations and analysis) Overall optimality based adaptive framework

Pushkin Kachroo(Ph.D, P.E., UNLV), Puneet Lakhanpal(MS, UNLV), Feedback MashaTheory Wilsonbased (Ph.D. Traffic OTS,Safety Nevada Campaigns