RHODES

RHODES:Fundamental Principles

Larry Head, Gardner Systems

Pitu Mirchandani, University of Arizona

TRB Annual Meeting 2000Workshop on Adaptive Signal Control Systems

RHODES

Overview

• Basic Philosophy of RHODES• Control Variables• Data Sampling, Filtering and Smoothing• Phasing Flexibility• Measures of Effectiveness• Oversaturated Conditions• Preemption/Priority

RHODES

Basic Philosophy of RHODES

• to proactively respond to and utilize the natural stochastic variations in traffic flow with the appropriate time scale

• to operate within the framework of North American traffic signal controllers

RHODES

System Architecture - Hierarchical

Destinations/Origins

Network LoadControl

Network FlowControl

IntersectionControl

Traffic SignalActivation

Detectors and Surveillance

Actual Travel Behavior and Traffic

NetworkLoads

Target Timings

ActualTimings

ControlSignal

Vehicle Flow Prediction

Scenario

Origins/Destinations

Current Capacities, Travel Times,Network Disruptions

(seconds)

(minutes)

(minutes/hours/days)

Platoon Flow Prediction

Network LoadEstimator/Predictor

Network FlowEstimator/Predictor

Intersection FlowEstimator/Predictor

Measurements

y(t)

ATIS

Historical/Infrastructure Data

RHODES

Control Variables

• Structural (static)– Geometric Description of Network– Location/Type of Detectors

• Traffic Dynamics Parameters (adaptive)– Saturation Flow, Turning Proportions…

• Signal Control (scheduled)– Phasing, Minimum, Maximum, Pedestrian,….

• Optimization Parameters

(interpreted to mean: Model and User Supplied Parameters)

RHODES

Control Variables

• Structural– Geometric Description

• Link-node representation

• Lanes, turning pockets, etc.

• Lane Channelization

• Lane Utilization

RHODES

Control Variables

• Structural– Detectors

• Location (e.g. 224’ upstream - Passage)

• Type– Passage (counting)

– Presence (stop bar)

– Detector Movement Assignment– Prediction Feed Assignment

RHODES

Control Variables• Traffic Dynamics Parameters

– Turning Percentage• Dynamic using OD Estimation (currently static)

– Queue Discharge Rates - by movement/phase• Saturation Flow Rate

• Dynamic using Queue Estimate and Presence Detectors

• Start-up Lost Time

– Link Free Flow Speed• Free Flow Corrected for Volume/Occupancy

RHODES

Control Variables

• Signal Control Parameters– Phase (optimization stage)

• Allowable movements

• Skipping

• Minimum Green

• Maximum Green (optional)

• Amber & Red Clearance Times

RHODES

Control Variables

• Optimization Parameters– Target Phase Evaluation Order

• ABCDACDE...

– Horizon• User-definable, now using 45 seconds

– Resolution• 1 second, 2 second, etc…….

RHODES

Data Sampling, Filtering and Smoothing (data characteristics)• Data Sampling

– data resolution = 1/second– detector signals

• passage (count of falling edges)

• presence (state of detector just before end of second)

– Signal state (phase, interval)

• Filtering = NONE

• Smoothing = NONE

RHODES

Phasing Flexibility

• Number of Phases– Any number of stages

• Flexibility in Phase Order– for any optimization - select desired phase order

A B C D E B

A B D B E C

RHODES

Phasing Flexibility

• Currently assumes a fixed phase order– rolling horizon =ABCDEA, BCDEAB,…..

• Phase Skipping allowed– user selectable– decisions = {0, min, min+1, ….., max*}

• for each phase

*optional

RHODES

Measures of Effectiveness

• Internal to RHODES– Queue Size (number of vehicles) Estimate– Predicted Link Flow Profiles– Predicted Delay

• based on current queue and predicted arrivals

• External– Queue Size – Predicted Arrival Profile

RHODES

Special Features for Oversaturated Conditions

• Consideration for Queue Spillback– adjust departure rate for movements with

upstream queue spillback– delay “discounting” for movements where

excessive downstream delay will occur

RHODES

Transit Priority

• Used coordination method – Progression band

• Priority Band for Detected Buses– Near upstream detection– Far side stations– Conditional on headway lateness

RHODES

Fire Priority

• Not in current RHODES model

• Potential route priority– Using coordination method (bandwidth)

• Preemption provided by underlying controller logic – Ignore the adaptive control commands when a

preemption event is timing

RHODES

Arterial/Network

• Designed for both

• Most experience/experimentation on arterials

• Optimization horizon (approx. 45 secs.)– need to populate predictions– travel time between intersections defines the

horizon over which optimization has data

RHODES

END SESSION 1

RHODES

RHODES:Equipment Requirements

Larry Head, Gardner Systems

Gary Duncan, Econolite Control Products

RHODES

Overview System Architecture Data requirements Communication requirements Local Controllers Central Hardware requirements Installation cost ranges Operations & Maintenance cost ranges

RHODES

Architecture

• RHODES–Hierarchical

• Network Loading

• Network Flow Control

• Management System

–Distributed• Local Intersection Control

RHODES

System Architecture

ATMS

Servers

Workstation

Workstation

Field Communications

LAN

RHODES

Architecture• Traffic Adaptive Signal Control is an added

capability of an ATMS• RHODES has been designed to operate as an

extension of existing ATMS systems• Requirements are for

– Additional Communications– Additional Detection– Additional Processing

RHODES

Architecture

ATMS

Servers

Workstation

Workstation

Field Communications

LAN

Additional Detection

Additional Communications

Additional Processor

**

*

RHODES

Data Requirements(Number, Type and Location of Sensors)

• Observability– need to be able to observe vehicle flows and

flow dynamics• Predictability

– need to be able to predict vehicle flows over a prediction horizon of interest

• Flexibility– need to accommodate wide range of detector

locations

RHODES

Data Requirements(Number, Type and Location of Sensors)

(i) (ii)

(iii)

(iv)

di

dA

di

dA

di

dA

di

dA

B B

B B

t t

t t

RHODES

Data Requirements(Number, Type and Location of Sensors)

Prediction Generators

Prediction Generators

Prediction Generators

PredictionReceiver

RHODES

Data Requirements(Number, Type and Location of Sensors)

• Detectors– Passage (upstream)

• Count of number of passed vehicles (trailing edge)

• Used for flow prediction and queue estimation

– Presence (stop bar)• State of detector at end of second

• Used for queue estimation– IF(presence = 0) queue =0

RHODES

Communications Requirements(Architecture, Polling Time, Bandwidth)

• Architecture– Peer-to-Peer Communications– Central-Intersection Communications

• Polling Time (options)– Discrete Event– 1 message/second

• Bandwidth– depends on architecture

RHODES

Communications Requirements(Architecture, Polling Time, Bandwidth)

• Architecture - Alternatives– Token Ring– Ethernet– Point-to-Point, Tree

• Technology– Field Hardened– $$$

RHODES

Peer-to-Peer Communications: Tucson and Seattle

Central Management System

Communications Hub

Intersection Controllers(2070 with MEN CPU)

Optelecom 9712 Modem Pairs:3 - 9600 Baud RS 2321 - 19.2K Serial/PTZ1 - Full Motion Video1 - 1200 Baud Voice

Figerlign Mux:1 - T1 Data (6 - RS-232)9 - Full Motion Video

All Communications are over single mode fiber.The link between the communications hub and the central management system in Seattle is TBD.

Figure 1: Peer-to-Peer Communications Architecture for Tucson and Seattle.

RHODES

Communications Requirements(Architecture, Polling Time, Bandwidth)

• Bandwidth (approx.)– central = 150 bytes/sec

• hub-central (8 intersections) = 1200 *10 (bits/byte)

• total = 12,000 bps– peer-to-peer packet:

• overhead (header, trailer, etc) 10 bytes• data (predictions, signal) + 30 bytes• packet (estimated total) = 40 bytes• 4 packets/sec = 160 bytes/sec• central + 150 bytes/sec• total = 310 *10 = 3100

– Recommend 19.2Kbps for Central, – 9600 bps for peer-to-peer

RHODES

Hardware Requirements(Central, Intermediate Field, Local Processor)

• Central– PC-based traffic server (e.g. iconsTM)– Serial Communications (e.g. Rocket Port)

• Intermediate Field– Field Hardened PC *+ Serial Comm

• Local Processor (options)– 2070 with VME Co-processor– standard controller+Co-processor

RHODES

Installation Cost Ranges

• Difficult to estimate– Project dependent– Architecture dependent

• Several projects estimated in the range of $45,000 - $50,000 per intersection including hardware + engineering

• License: The University of Arizona (approx. $500/intersection)

RHODES

O&M Cost Ranges

• Incremental cost based on additional hardware (including detection) and software

• Cost savings based on improved signal timing