VII California Team Meeting

1
VII California Team Meeting

• November 2, 2006
• Metropolitan Transportation Commission

2

• AGENDA
• Emerging Research Results
• Fused Detection
• Curve Overspeed
• RSE Status
• Intersection Sniffer
• Backhaul - T1, 3G, WiMax
• WiFi
• APTA Demo and Beyond
• ITS America Demo

3
Emerging Research Results
4
Fused Traffic Detection for Arterials
5
Objective

• Fuse traffic detection data
• Loop detector data (count, occupancy)
• Traffic signal status data (phase duration, phase
  interval, pedestrian calls)
• VII probe vehicle data (speed, travel time,
  trajectory)
• Other data (e.g. toll tag)
• Measure arterial performance
• Travel time
• Queue length
• Intersection delay
• Number of vehicle stops

6
Existing Data Fusion Approach

• Artificial neural network (ANN)
• Pro capable of giving correct result in the
  presence of noisy or missing data
• Con ANN is not a systematical approach, so the
  results rely on model training
• Score ranking for different sources and pick up
  the best
• Primarily rely on loop detector data and use
  probe vehicle data as enhancements
• Statistic approach (Dempster-Shafer evidence
  theory)

7
Work Plan

• Model development
• Development of models to pre-process raw data
• Data fusion
• Approach I VII probe data for model calibration
  purpose

VII probe data
Calculated travel time
P
Loop detector data
LS
Model calibration
Predicted travel time
Predicted travel time
Signal status data
Other data
Calculated travel time
O
8
Work Plan (cont.)

• Approach II ANN
• Model evaluation
• Simulation analysis
• Field data analysis

Loop detector data
LS
Predicted travel time
Signal status data
ANN
VII probe data
Calculated travel time
Predicted travel time
P
Other Data
Calculated travel time
O
9
Simulation Analysis Plan

• Get familiar with VISSIM
• Simulation environment
• Network construction
• Simulation functions
• Programming interface
• Component Object Model (COM)
• Data collection module
• Signal control module

10
Simulation Analysis Plan

• Data collection
• Loop detector data
• Signal status data
• VII probe vehicle data
• Other data
• Coding the pre-process models using API
• Evaluation of pre-process models
• Coding and evaluation of two data fusion
  approaches

11
Progress To Date

• Model development based on loop and signal status
  data
• Constructing simulation network in VISSIM (El
  Camino Real in Palo Alto)
• Warm-up on simulation API
• Simulation parameters control
• Automatic vehicle travel time capture
• Average travel time statistics for selective
  vehicles
• Simple display interface
• Adjustment of probe vehicle penetration rate

12
Progress To Date
13
Curve Overspeed Warning
14
Review of Curve Over-Speed Warning Research

• Map data extraction for developing application
  map.
• Dynamic road curvature estimation using digital
  map data.
• OBU sensor configuration and sensor data fusion.
• Warning algorithm design.

15
Status

• Progress
• Conventional Spline approach is not favorable for
  our application due to accuracy and robustness
  problems.
• We have come up with a new approach -
  Circle-Center Searching algorithm to address the
  two problems.
• New approach has been tested on three types of
  curves
• Arc
• Circle
• S-Curve
• Specify on-board sensors (sent to Toyota ITC and
  VWERL)
• On-Going Work
• Setting up the 1st OBU at PATH
• Extract more map data to validate our new
  curvature estimation algorithm
• Conduct preliminary field tests for warning
  algorithm development with Toyota and VW

16
Why is Spline approach unfavorable?
17
Circle-Center Searching Algorithm

• Concept
• Geometrically, vehicles trajectory is consisted
  of moving circles.
• Road can be modeled by combination of circles
• Method
• Treat road as circles
• Pre-estimate radius and circle center location
• Determine center location, radius, and curve type
• Adaptively estimates of radius and curvature
  center
• Reiterate if required
• Advantage
• Simple computation
• Accurate
• Robust

18
Spline Approach vs. Circle-Center Searching
Approach
(Part of Bayview Ave).
Circle-Center Searching Approach (New)
Our new approach is able to (1) achieve more
accurate and stable curvature estimates (2)
locate the center of turning radius and (3)
detect the change of curvature.
19
Example 1 Arc with Curvature Change Curve on
Bayview Ave.
Verified by Google Earth and test drive.
20
Example 2 Circle On-Ramp from Marsh Rd. to SB101
Verified by Google Earth
21
Example 3 S-Curve Off-Ramp from SB101 to Marsh
Rd.
22
On-Going Work

• Extract more map data of highway ramps to
  validate our new curvature estimation algorithm.
• Conduct preliminary field tests for warning
  algorithm development.
• Refine vehicle dynamics analysis and include more
  complex road attributes.

• One GPS test at PATH

• Setup a RSU at PATH for preliminary tests.

23
RSE
24
10 1 RSUs
25
CICAS V RSU

• 5th and El Camino Real in Atherton
• Site was surveyed by Daimler-Chrysler, Caltrans,
  and PATH few weeks ago
• New RSU design (Technocom) using WIFI instead of
  fiber
• RSU sits inside a weather proof enclosure outside
  the cabinet attached to the vertical signal pole
• WRM antenna on the mast arm
• Backhaul ????
• Final design must be approved by Caltrans before
  installation (by both Mr. Price and Mr. Chiu)
• Possibly in direct line of site to one of WiMax
  antennas set up for Caltrain (by Intel)

26
(No Transcript)
27
(No Transcript)
28
VII in California

• Procurement
• Cost Minimization (Borrow?, Steal?, Donation?)
• Charged to the existing VII contract
• Coordination
• D4 and the rest of the gang
• Installation (5th and El Camino Real)
• Field inspection
• Final Installation

29
Sniffer
30
Development ActivitiesSniffer Broadcast
31
Signal Phase and Intersection Message Set
Broadcast

• Signal sniffer hardware and intersection message
  set broadcast software development
• APTA
• PATH_at_20
• Broadcast software
• originally developed using messages sent over
  Ethernet from the Econolite controller
• Adapted to use signal phase information from
• NTCIP and AB3418 protocols
• Sniffer
• Future changes require message formatting code to
  be well-separated from the details of phase
  tracking
• e.g., separating dynamic signal phase tracking
  information from static geographical information.

32
Signal Sniffer Block Diagram
See next slide for whats in this block
The GREEN signal for each approach is being
sensed by the sniffer circuits that we have
installed so far.
33
PATH Signal Phase Tracking and Broadcast Software
Block Diagram
Alternative software modules provide uniform
interface to phase tracker.
Separate module formats message and controls
broadcast frequency.
Tracks count-down based on timing plan and
dynamic signal phase inputs
Digital I/O from sniffer circuit
DB_ATSC
Phase Tracking Module
DB_IX
170 or 2070 Serial Port
Configuration files with timing plan and
geographic (approach and stop line) information
Currently broadcast as UDP message over Denso WRM
34
Intersection Message Broadcast tested with
sniffer at two locations

• Page Mill/El Camino
• actuated/coordinated

• Richmond Field Station
• fixed timing

35
Sniffer testing so far

• Performance
• Reliable, with minimal delay in reporting signal
  state changes
• Broadcasts accurate countdown for yellow period,
  as planned.
• At Page Mill/El Camino actuated coordinated
  intersection, best effort broadcasts based on
  timing plan for red and green periods shows some
  ending a few seconds earlier than predicted, some
  lasting a few seconds longer.

36
Issues raised during sniffer testing at RFS

• Broadcasts of 10 messages per second caused the
  application to show increasing delays, while
  broadcasts at 5 messages per second were fine
• Could be due to interference from other traffic
  at sending RSU
• Could be due to overload from on-board
  applications on Windows
• Requires further investigation
• Observed clock jitter
• On the order of hundreds of milliseconds over the
  course of a 45 second phase was observed between
  the PC104 doing the phase tracking and the 2070
  signal controller
• clock corrections important, through GPS or other
  means

37
Intersection Message Broadcast Work in Progress

• Intersection Message Set send/receive testing
  software made available to partners this week
• 2070 with Caltrans Traffic Signal Control Program
  (TSCP) delivered to PATH last week for AB3418
  protocol work
• Need to determine by testing with serial port to
  Linux PC104 how fast status can be reported
• Need to discuss contents of LongStatus8 message
  with car company partners to see if any fields
  ought to be added to the intersection message set
• Need to investigate hardware to forward serial
  port information over wireless IP to Technocom
  RSU
• NTCIP code that runs on QNX and communicate with
  Siemens 2070 software needs to be ported to Linux
  and compared directly with sniffer performance

38
Backhaul T1
We have requested T1 installations at 2 remaining
sites. One is through SpeakEasy and one with ATT.

• Marsh/US 101 (Menlo Park)
• Route 85/US 101 (Mountain View)

Lessons Learned The cabinet location/addresses
are nearly impossible to use for determining the
existing circuits that lead into the cabinets.
39
Backhaul - 3G
9 Verizon Raven-E Modems have been deployed.
The AirCard 875 is about to be delivered and is
supposed to provide up to 3.6 Mbps download.
40
Backhaul - WiMAX
Vince was able to review the candidate sites that
you sent.  There are a few likely candidates.  He
is actually in the area all week so plans on
doing a site survey of these sites to determine
feasibility.  I will let you know what he comes
back with
41
WiFi
42
WiFi for VII CA?

• 1 SC-101-397 101 Santa Clara South Airport
  Parkway/Brokaw
• 2 SC-101-522 101 Santa Clara North Oregon/Embarcad
  ero
• 3 SM-101-112 101 San Mateo North Hillsdale
• 4 SC-101-332 101 Santa Clara North Tully
• 5 SM-101-187 101 San Mateo South SFO Terminal
• 6 SC-280-156 280 Santa Clara North El Monte
• 7 SC-280-18 280 Santa Clara North 10th Street
• 8 SM-101-262 101 San Mateo North Harney Way

• Two Tracks
• Eight Dash (Circum Nav) WiFi Fixed Access Points
• Build Two Additional WiFi RSE Near San Antonio /
  101

43
Initial WiFi RSE Design

• Next Steps
• Coordinate Backhaul
• Schedule D4 Review
• Build / Test
• Schedule Installation _at_ Two TBD Sites
• Questions
• Would OEMs Develop Applications and Use?
• Overcome by Events? (Dash WiFi RSU)

44
APTA Demo and Beyond
45
VII Transit Demo

• October 9th Fuel Problem
• VII Functions Developed
• Travel time and incident updates
• Sniffer demo at El Camino and Pagemill
• Caltrain schedule display near Caltrain station
• 511.org information transmitted via wireless
  internet into bus

46
APTA Demo and Beyond

• Probe/travel time application worked in demos
• However, was limited to predefined OD pair
• No GPS, so vehicle and server assume travel is
  between predefined RSU origin and destination
• But, we can encode GPS in beacon message
• For now, data comes from google maps
• App works for any RSU OD pair (or sequence)
• Generates travel time message for the origin RSU
• Can run demos without extra preparation
• Also works if car does have GPS

47
Probe Details

• Sample C code in the c-api download
• probe-vehicle.c
• Can use viicatl executable
• probe-vehicle and probe-server modes
• Probe server is at
• path.berkeley.edu10080
• Vehicle code selects probe server by ipaddr
• Suggestion for new device status (in snapshot)
• field name GPS status
• field value number of satellites (no GPS gt 0)

48
Temporary Signage Messages

• We now have a telnet interface
• Allows you to add signage messages interactively
• Easy to read text format
• Quick-and-dirty alternative to writing a signage
  generator using the VII-CA binary format
• You can specify
• Metadata rsu ID, message ID, expiration
  (minutes)
• Data text, sounds, driving time, destination,
  ...
• Message is at RSU within 1 minute

49
Sample data entry

• event_type 38
• event_subtype 0
• longitude 0.0
• latitude 0.0
• heading 0.0
• distance 0.0
• start_time 315 AM, October 25, 2006
• end_time Thurs, Oct 26, 2006, 1800 UTC
• link_speed 65
• truck_speed 55
• speed_limit_road_name US-101
• destination_name San Jose
• driving_time 0
• text Your message here!
• sounds foo.wav,bar.wav,baz.wav

50
Temporary Signage Details

• Telnet server is at
• path.berkeley.edu10081
• Not the same as the PATH signage server
• Contact us for password
• Tell us your IP address (for firewall)
• Documentation is printed at start of session
• Each partner group has its own namespace
• No conflicts between partners
• But signage is visible to all users

51
ITSA and VII California
52
Particulars

• VII Pavilion has accompanying demos
• VII California has reserved a space
• Demo concept
• Travel time with mobile RSE / in-vehicle signage
• TBD vehicle platform for OBE
• TBD probe and data source concept
• TBD whether to implement RSE on Caltrans ROW
• Issues
• Need to scout location
• Someone gets to travel to Palm Springs!
• Palm Springs in the Summer (4 6 June 07)
• Hot for people, hot for equipment
• Distance and environment means travel for week(s)
  x people
• Weekly coordination telecons have begun (Tues, 9
  10 am) and need staffing
• Questions
• If / who / what / how to do this demo?