this traffic signal will turn green? Why do I want to know when - - PowerPoint PPT Presentation

▶
this traffic signal will
SMART_READER_LITE
LIVE PREVIEW

this traffic signal will turn green? Why do I want to know when - - PowerPoint PPT Presentation

Jan 21, 2024 299 likes •802 views

How do I know, when this traffic signal will turn green? Why do I want to know when the signal turns green? 28.03.2012 Seminar in Distributed Computing Gianin Basler Introduction Traffic light countdown timer 28.03.2012 Seminar in

slide-1
SLIDE 1

28.03.2012 Seminar in Distributed Computing Gianin Basler

How do I know, when this traffic signal will turn green? Why do I want to know when the signal turns green?

slide-2
SLIDE 2

28.03.2012 Seminar in Distributed Computing Gianin Basler

Introduction

Traffic light countdown timer

slide-3
SLIDE 3

28.03.2012 Seminar in Distributed Computing Gianin Basler

Introduction

Traffic light countdown timer

slide-4
SLIDE 4

28.03.2012 Seminar in Distributed Computing Gianin Basler

Introduction

Traffic light countdown timer

  • Expensive
  • Impractical deployment
  • Costly maintenance
slide-5
SLIDE 5

28.03.2012 Seminar in Distributed Computing Gianin Basler

Introduction

SignalGuru Joint project of Princeton University and MIT Demonstrates potential of smartphone cameras Presented at MobiSys’11

slide-6
SLIDE 6

28.03.2012 Seminar in Distributed Computing Gianin Basler

Introduction

SignalGuru Basic idea

  • Take picture of intersection
  • Filter out relevant traffic signal
  • Predict the next green phase

Advantages

  • No infrastructure
  • Runs on mobile phones
  • Detects and predicts traffic signals
slide-7
SLIDE 7

28.03.2012 Seminar in Distributed Computing Gianin Basler

Outline

  • 1. Traffic Light Background
  • 2. SignalGuru
  • 3. Applications
  • 4. Related Work
slide-8
SLIDE 8

28.03.2012 Seminar in Distributed Computing Gianin Basler

Traffic Light Background

Terminology

  • Phase:

different, but non-conflicting movements

  • Cycle:

each phase had green once

  • Phase length: green light duration for a phase
  • Cycle length:

sum of all phase lengths

North South East West

1. Traffic Light Background 2. SignalGuru 3. Applications 4. Related Work

slide-9
SLIDE 9

28.03.2012 Seminar in Distributed Computing Gianin Basler

Traffic Light Background

2 types of traffic lights Pre-timed

  • Settings (i.e. phase and cycle lengths) are fixed
  • Same schedule repeats every cycle
  • Typically 3 modes of operation

Adaptive

  • Uses inductive loop detectors
  • Adjusts settings based on lane saturation
  • Changes settings every cycle
  • Phases scheduled in deterministic, round-robin manner

1. Traffic Light Background 2. SignalGuru 3. Applications 4. Related Work

slide-10
SLIDE 10

28.03.2012 Seminar in Distributed Computing Gianin Basler

Outline

  • 1. Traffic Light Background
  • 2. SignalGuru

a) Modules b) Challenges

  • 3. Applications
  • 4. Related Work

How do I know, when the traffic signal will turn green?

slide-11
SLIDE 11

28.03.2012 Seminar in Distributed Computing Gianin Basler

SignalGuru - Modules

Detection module Collaboration module Transition filtering module Prediction module

1. Traffic Light Background 2. SignalGuru 3. Applications 4. Related Work

slide-12
SLIDE 12

28.03.2012 Seminar in Distributed Computing Gianin Basler

SignalGuru - Detection

Setup Windshield mounted iPhones Phone cameras capture video frames Detection activated based on GPS location Processes a new frame every 2 seconds

2. SignalGuru Modules

  • Detection
  • Transition Filtering
  • Collaboration
  • Prediction
slide-13
SLIDE 13

28.03.2012 Seminar in Distributed Computing Gianin Basler

SignalGuru - Detection

Characteristics of a traffic light

  • Bright bulb colour
  • Bulb shape (circle, arrow)
  • Black traffic signal housing
  • High above ground

2. SignalGuru Modules

  • Detection
  • Transition Filtering
  • Collaboration
  • Prediction
slide-14
SLIDE 14

28.03.2012 Seminar in Distributed Computing Gianin Basler

Colour filter

SignalGuru - Detection

2. SignalGuru Modules

  • Detection
  • Transition Filtering
  • Collaboration
  • Prediction
slide-15
SLIDE 15

28.03.2012 Seminar in Distributed Computing Gianin Basler

SignalGuru - Detection

slide-16
SLIDE 16

28.03.2012 Seminar in Distributed Computing Gianin Basler

SignalGuru - Detection

Colour filter Laplace edge detection

2. SignalGuru Modules

  • Detection
  • Transition Filtering
  • Collaboration
  • Prediction
slide-17
SLIDE 17

28.03.2012 Seminar in Distributed Computing Gianin Basler

SignalGuru - Detection

slide-18
SLIDE 18

28.03.2012 Seminar in Distributed Computing Gianin Basler

SignalGuru - Detection

Colour filter Hough transform Laplace edge detection

2. SignalGuru Modules

  • Detection
  • Transition Filtering
  • Collaboration
  • Prediction
slide-19
SLIDE 19

28.03.2012 Seminar in Distributed Computing Gianin Basler

SignalGuru - Detection

10 9 10 4 4 3 4 2 9

slide-20
SLIDE 20

28.03.2012 Seminar in Distributed Computing Gianin Basler

SignalGuru - Detection

Colour filter Calculate BCC and BBC Hough transform Laplace edge detection BCC ∗ BBC > threshold?

slide-21
SLIDE 21

28.03.2012 Seminar in Distributed Computing Gianin Basler

SignalGuru - Detection

10 9 10 4 4 3 4 2 9

BCC = Bulb Colour Confidence Is the object in correct colour range? BBC = Black Box Confidence Is the object surrounded by a traffic signal housing? 0.95 0.2

2. SignalGuru Modules

  • Detection
  • Transition Filtering
  • Collaboration
  • Prediction

0.6 0.15

slide-22
SLIDE 22

28.03.2012 Seminar in Distributed Computing Gianin Basler

SignalGuru - Detection

Colour filter Calculate BCC and BBC Hough transform Report no traffic light found Report traffic light (colour, centre coordinates, radius) Laplace edge detection BCC * BBC > threshold?

2. SignalGuru Modules

  • Detection
  • Transition Filtering
  • Collaboration
  • Prediction
slide-23
SLIDE 23

28.03.2012 Seminar in Distributed Computing Gianin Basler

Outline

  • 1. Traffic Light Background
  • 2. SignalGuru

a) Modules b) Challenges

  • 3. Applications
  • 4. Related Work
slide-24
SLIDE 24

28.03.2012 Seminar in Distributed Computing Gianin Basler

SignalGuru - Challenges

How to run everything with limited processing power? Make use of high placement of traffic signals Reduce detection window size Benefits: a) Processing time decreased by 41% (from 1.73s to 1.02s) b) Almost halves misdetection rate (from 15.4% to 7.8%)

3. SignalGuru Challenges

  • Processing Power
  • Ambient Light Conditions
slide-25
SLIDE 25

28.03.2012 Seminar in Distributed Computing Gianin Basler

SignalGuru - Challenges

How to run everything with limited processing power? Detection window

3. SignalGuru Challenges

  • Processing Power
  • Ambient Light Conditions
slide-26
SLIDE 26

28.03.2012 Seminar in Distributed Computing Gianin Basler

SignalGuru - Challenges

How to deal with variable ambient light conditions? LED traffic signals have fixed intensity Adjust and lock camera exposure time

3. SignalGuru Challenges

  • Processing Power
  • Ambient Light Conditions
slide-27
SLIDE 27

28.03.2012 Seminar in Distributed Computing Gianin Basler

SignalGuru - Detection in action

slide-28
SLIDE 28

28.03.2012 Seminar in Distributed Computing Gianin Basler

SignalGuru - Detection

2. SignalGuru Modules

  • Detection
  • Transition Filtering
  • Collaboration
  • Prediction

Summary Phone camera captures video frames Algorithm filters out relevant traffic light Reports location, radius and colour of a detected traffic light Red x:4.05, y: 3.22 r: 0.05 Signal will turn green in 24s

?

slide-29
SLIDE 29

28.03.2012 Seminar in Distributed Computing Gianin Basler

Outline

  • 1. Traffic Light Background
  • 2. SignalGuru

a) Modules b) Challenges

  • 3. Applications
  • 4. Related Work

How do I know, when the traffic signal will turn green?

slide-30
SLIDE 30

28.03.2012 Seminar in Distributed Computing Gianin Basler

SignalGuru - Transition Filtering

Detection module’s output is fairly noisy While waiting at traffic light: 65% false transition detection Need to filter out false positives

2. SignalGuru Modules

  • Detection
  • Transition Filtering
  • Collaboration
  • Prediction
slide-31
SLIDE 31

28.03.2012 Seminar in Distributed Computing Gianin Basler

SignalGuru - Transition Filtering

Two-stage filter Low pass filter 88% of false positives in single frame Colocation filter Red and green bulb contained in the same black box

frame i + 1 frame i

2. SignalGuru Modules

  • Detection
  • Transition Filtering
  • Collaboration
  • Prediction
slide-32
SLIDE 32

28.03.2012 Seminar in Distributed Computing Gianin Basler

SignalGuru - Collaboration

Exchange time stamped R -> G transitions Use ad-hoc 802.11g network connection The more transition data, the more accurate the prediction.

2. SignalGuru Modules

  • Detection
  • Transition Filtering
  • Collaboration
  • Prediction
slide-33
SLIDE 33

28.03.2012 Seminar in Distributed Computing Gianin Basler

SignalGuru - Prediction

Pre-timed traffic signals Main challenge: Accurately synchronise SignalGuru’s clock with phase transition How it’s done: Achieved by capturing a colour transition Rest of the data available from traffic authorities

2. SignalGuru Modules

  • Detection
  • Transition Filtering
  • Collaboration
  • Prediction
slide-34
SLIDE 34

28.03.2012 Seminar in Distributed Computing Gianin Basler

SignalGuru - Prediction

Traffic signal timeline 𝑢 = detected signals and transitions 𝑄𝑀 = phase length 𝜐 = predicted transitions 𝜁 = error

2. SignalGuru Modules

  • Detection
  • Transition Filtering
  • Collaboration
  • Prediction

B B A A

𝑢𝐵,𝑆→𝐻 𝜐𝐶,𝑆→𝐻 𝜐𝐵,𝑆→𝐻 𝑢𝐵,𝑆 𝑢𝐵,𝐻 𝑄𝑀𝐵 𝑄𝑀𝐶 𝜁

B A

slide-35
SLIDE 35

28.03.2012 Seminar in Distributed Computing Gianin Basler

SignalGuru - Prediction

Adaptive traffic signals Main challenge: Predict the phase length How it’s done: Measure and collaboratively collect transition history Feed data to Support Vector Regression prediction model

2. SignalGuru Modules

  • Detection
  • Transition Filtering
  • Collaboration
  • Prediction
slide-36
SLIDE 36

28.03.2012 Seminar in Distributed Computing Gianin Basler

SignalGuru - Prediction

Support Vector Regression 2 phases:

  • 1. Training: create a prediction model (offline)

2. SignalGuru Modules

  • Detection
  • Transition Filtering
  • Collaboration
  • Prediction

SVR History data SVR Model Prediction scheme

slide-37
SLIDE 37

28.03.2012 Seminar in Distributed Computing Gianin Basler

SignalGuru - Prediction

Support Vector Regression 2 phases:

  • 1. Training: create a prediction model (offline)
  • 2. Prediction: predict next phase length

2. SignalGuru Modules

  • Detection
  • Transition Filtering
  • Collaboration
  • Prediction

Current data SVR Model Prediction scheme Next phase length

slide-38
SLIDE 38

28.03.2012 Seminar in Distributed Computing Gianin Basler

SignalGuru - Prediction

Support Vector Regression Prediction schemes PS1: Prediction based on history for the same phase PS2: Also use lengths of preceding phases in same cycle PS3: Use data of the last 5 cycles

2. SignalGuru Modules

  • Detection
  • Transition Filtering
  • Collaboration
  • Prediction
slide-39
SLIDE 39

28.03.2012 Seminar in Distributed Computing Gianin Basler

Outline

  • 1. Traffic Light Background
  • 2. SignalGuru
  • 3. Applications
  • 4. Related Work

Why do I want to know when a signal turns green?

slide-40
SLIDE 40

28.03.2012 Seminar in Distributed Computing Gianin Basler

Applications - GLOSA

Green Light Optimal Speed Advisory Advise drivers on optimal speed Avoid stopping at red light Benefits a) Decreases fuel consumption by 20% b) Smoothens and increases traffic flow c) Decreases environmental impact

3. SignalGuru Applications

  • GLOSA
  • TSAN
slide-41
SLIDE 41

28.03.2012 Seminar in Distributed Computing Gianin Basler

Applications - GLOSA

SignalGuru’s GLOSA screen

3. SignalGuru Applications

  • GLOSA
  • TSAN
slide-42
SLIDE 42

28.03.2012 Seminar in Distributed Computing Gianin Basler

Applications - TSAN

Traffic Signal-Adaptive Navigation Avoid long waits at red lights Advise drivers on possible detours Benefits a) No stops at red lights b) Reduces travel time

3. SignalGuru Applications

  • GLOSA
  • TSAN
slide-43
SLIDE 43

28.03.2012 Seminar in Distributed Computing Gianin Basler

Outline

  • 1. Traffic Light Background
  • 2. SignalGuru
  • 3. Applications
  • 4. Related Work
slide-44
SLIDE 44

28.03.2012 Seminar in Distributed Computing Gianin Basler

Related Work

Hazardous Location Warning Vehicle detects hazardous location, i.e. oil spill Transmits data to oncoming vehicles Makes use of

  • Car sensors
  • Ad-hoc network

Source: http://www.car-to-car.org/index.php?id=196

4. Related Work

  • Location Warning
  • ParkNet
slide-45
SLIDE 45

28.03.2012 Seminar in Distributed Computing Gianin Basler

Source: http://www.winlab.rutgers.edu/~gruteser/papers/mathur_parknet10.pdf

Related Work

ParkNet Drive-by Sensing of Road-Side Parking Statistics Project of Rutgers University, USA Issue Searching for parking spot creates congestion Lead to a loss of $78 billion in 2007 in US

  • 4.2 billion lost hours
  • 11 billion litres of wasted fuel

4. Related Work

  • Location Warning
  • ParkNet
slide-46
SLIDE 46

28.03.2012 Seminar in Distributed Computing Gianin Basler

Source: http://www.winlab.rutgers.edu/~gruteser/papers/mathur_parknet10.pdf

Related Work

ParkNet Drive-by Sensing of Road-Side Parking Statistics Mobile system with sensors on cars Ultrasonic sensor and GPS receiver

4. Related Work

  • Location Warning
  • ParkNet
slide-47
SLIDE 47

28.03.2012 Seminar in Distributed Computing Gianin Basler

Related Work

ParkNet Data uploaded using Wi-Fi Central server creates parking map

4. Related Work

  • Location Warning
  • ParkNet
slide-48
SLIDE 48

28.03.2012 Seminar in Distributed Computing Gianin Basler

Related Work

ParkNet Allows checking of near-real-time parking situation Eliminates need to search for parking Benefits a) Saves time b) Saves a lot of fuel

4. Related Work

  • Location Warning
  • ParkNet
slide-49
SLIDE 49

28.03.2012 Seminar in Distributed Computing Gianin Basler

The End

Questions?

Thank you for your attention!