POLARIS - Planning and Operations Language for Agent- based Regional Integrated Simulation

POLARIS: Planning and Operations Language for Agent-based Regional Integrated Simulation

Dr. Kuilin Zhang [email protected]

Department of Civil and Environmental Engineering

Michigan Technological University Houghton, MI

Motivation

• The existing implementation gap between activity-based modeling (ABM) of travel demand and simulation-based dynamic traffic assignment (DTA) modeling of transportation network

• Not truly integrated: mostly existing implementations are decoupled using files as interface between ABM and DTA

• The knowledge barrier: Researchers from either demand or network side may simplify the other side models

• The use of the emerging high-performance parallel computing technologies in the hybrid multi-core CPU and many-core GPU systems

• Running 24-hour and second-by-second resolution of millions of travelers’ multidimensional travel decisions in large-scale metropolitan areas in a workstation or powerful laptop computer

• Can support real-time applications to support real-time traffic management center (TMC) operational strategies for uncertainty events by feeding real-time and historical traffic observations, traffic events, and traffic estimation and predictions

• An open source agent-based software design framework provides a new concept for performance and re-usability

POLARIS (Planning and Operations Language for Agent-based Regional Integrated Simulation)

• Sponsored by FHWA, USDOT, 2011-2013 • Model Traffic Management Centers and other ITS Systems • Enhance Interoperability among Demand, Network, and Operation Models

• Research Team – TRACC, Argonne National Laboratory • Hubert Ley (manager), Mike Hope (computing and visualization), Vadim Sokolov

(data interface and operation), Josh Auld (demand), Bo Xu (computing, now at HERE, Nokia), and Kuilin Zhang (network, now at Michigan Tech)

• Fundamental Goals and Philosophies • Develop transportation modeling implementation standards and protocols • Create an open source model development environment • Connect researchers from demand, network, and operation communities with a

common modeling language • Offer high-performance computing tools while maintaining flexibility and

modularity

POLARIS Repository Structure

Core Library

Interprocess Communication

Discrete Event Engine

POLARIS Meta-Structures

Memory Allocator

Open Source Versioned Repository

Reusable Transportation

Interfaces

Fundamental Transportation Data Layouts

Final User Application

Modular Transportation

Algorithms

Open Source “Playground”

Specialized Transportation Data Layouts

Experimental Transportation

Algorithms

Computing Researchers

Modelers and Engineers

Modeling Researchers

An Agent-based Approach

Element

Entity Faculty

Directable Agent

Routable Agent

Some Destination Chooser

Some Destination Chooser

Some Route Chooser

Get Route ( ) Basic

Network Path

Route Chooser

Network Path

Fast Route Chooser

Arrival Variance

Network Path

Accurate Route Chooser

Est. Travel Time

Get Dest. ( ) Basic

Network Dest.

Destination Chooser

Network Dest.

Fast Destination Chooser

Suitability

Network Dest.

Robust Dest. Chooser

Alternative Dest.

Get Dest. ( ) Fast

Get Dest. ( ) Robust

Get Route ( ) Basic

Get Route ( ) Accurate

Get Route ( ) Basic

Get Route ( ) Fast

Get Dest. ( ) Basic

Get Dest. ( ) Basic

Requirements Capabilities

POLARIS Core Library: Re-Usable Low Level Capabilities

• Discrete Event Engine and Automated Multi-Threading • Enable code writing from an agent-based perspective

• Memory Allocation Library • Optimized for use in transportation modeling applications

• Thread-Safe Structures • Specialized for time-dependent nature of simulations

• Inter-Process Communication for use in Cluster Computing • Ease connection and communication of disparate processes

POLARIS • Integrated Dynamic Demand

and Network Modeling • Travel behavior: activity,

duration, location, departure time, mode, route, etc.

• Traffic Simulation • Driving behavior: car-following,

lane changing, etc. • Intelligent Transportation

Systems • Smart signals, ramp metering,

variables message signs, sensors, GPS, etc.

• Advanced Vehicles • Connected and automated

vehicles, electric vehicles, etc.

ITS Facilities: Variable Message Signs, Variable Speed Signs, Ramp Meters

Real-Time Events: Weather (e.g. Snow) and Accidents

Connected Vehicles

Driver-Centric On-Line Lane-Based Microscopic Traffic Simulation and

Optimal Routing

Modeling Millions of Persons’ Travel Decisions and Network Traffic Dynamics Second-by-Second in Metropolitan Areas

Individual Vehicle Driving Simulation • Vehicle Driving Simulation

• Kinematic Wave Traffic Flow Theory

• Mesoscopic Simulation • Microscopic Simulation

(potential) • Car-Following • Lane-Changing

• Advanced Vehicles (potential) • Connected Vehicles

• Connected vehicles through smartphones

• Connected vehicles through DSRC

• Vehicle-to-vehicle communication

• Vehicle-to-infrastructure communication

• Automated Vehicles • Electric Vehicles

Individual Traveler’s 24-hour Activity Chains

• An Activity Chain • Home – Service –

Social – Home • Three Trips • Three Locations

• Advanced Vehicles (potential)

• Electric Vehicles • Charging location

selection • Connected Vehicles

• Location-Based Services

Intelligent Vehicle Path Switching in response to Real-Time Traffic Information

• Intelligent Vehicles • Connected through

Smartphones • Radio • VMS

• Travel Behavior Change in response to Real-Time Traffic Information

• Change route • Change activity

location • Change activity

duration • Change departure

time • Change activity

schedule

Original Path

Current Path

Destination

Switching Point

An Intelligent Vehicle

Weather Events – Snow • Travel behavior Change in

response to Weather Information

• Change route • Change departing time • Change activity location • Change activity schedule

• Driving Behavior Change in response to Weather Impacts

• Change speed • Road capacity drop

• Traffic Management Center (TMC)

• How to disseminate weather information

• How to dispatch plow trucks

Accident Events • Travel Behavior

Change in response to Real-Time Accident Information

• Change route • Change activity

location

• Driving Behavior Change in response to Accident Events

• Capacity reduction • Accident event

duration prediction

• Traffic Management Center (TMC)

• How to disseminate real-time accident information

• How to dispatch two truck to accident location

An Accident on Lakeshore Dr

An Accident on I-290

Summary

• An open source transportation modeling development environment and language

• A set of core libraries provide memory efficient high-performance computing capabilities using an agent-based approach

• A 3D GUI for large-scale visualization and real-time interventions • Provide interfaces to model and simulate the travel and drive

decisions of the emerging advanced vehicle technologies such as connected, automated, and electrified vehicles

Thank You

Kuilin Zhang [email protected]