multi-agent control of a fleet of cybercarsmulti-agent control of a fleet of cybercars renshi luo...
TRANSCRIPT
Multi-Agent Control of A fleet of Cybercars
Renshi Luo
Delft Center for Systems and ControlDelft University of Technology
September 5, 2013
Renshi Luo (DCSC, TUD) 1 / 21
Who I am
Renshi Luo (DCSC, TUD) 2 / 21
Who I am
• second year Ph.D candicate
Renshi Luo (DCSC, TUD) 2 / 21
Who I am
• second year Ph.D candicate
• member of Hybrid and Distributed Systems and Control group
Renshi Luo (DCSC, TUD) 2 / 21
Who I am
• second year Ph.D candicate
• member of Hybrid and Distributed Systems and Control group
• work with Ton van den Boom and Bart De Schutter
Renshi Luo (DCSC, TUD) 2 / 21
Do you drive car?
Renshi Luo (DCSC, TUD) 3 / 21
Renshi Luo (DCSC, TUD) 4 / 21
• accidents
Renshi Luo (DCSC, TUD) 4 / 21
• accidents
• congestion
Renshi Luo (DCSC, TUD) 4 / 21
• accidents
• congestion
• energy consumption
Renshi Luo (DCSC, TUD) 4 / 21
• accidents
• congestion
• energy consumption
• pollution
Renshi Luo (DCSC, TUD) 4 / 21
Renshi Luo (DCSC, TUD) 5 / 21
Public transport systems
• considered suitable solutions to those problems
Renshi Luo (DCSC, TUD) 6 / 21
Public transport systems
• considered suitable solutions to those problems
• e.g., buses, trams, subways, etc
Renshi Luo (DCSC, TUD) 6 / 21
Public transport systems
• considered suitable solutions to those problems
• e.g., buses, trams, subways, etc
• widely used and contiuously improved
Renshi Luo (DCSC, TUD) 6 / 21
Public transport systems
• considered suitable solutions to those problems
• e.g., buses, trams, subways, etc
• widely used and contiuously improved
• * predefined schedules and routes
Renshi Luo (DCSC, TUD) 6 / 21
Public transport systems
• considered suitable solutions to those problems
• e.g., buses, trams, subways, etc
• widely used and contiuously improved
• * predefined schedules and routes
In terms of personal mobility, private cars still win !!
Renshi Luo (DCSC, TUD) 6 / 21
Public transport systems
• considered suitable solutions to those problems
• e.g., buses, trams, subways, etc
• widely used and contiuously improved
• * predefined schedules and routes
In terms of personal mobility, private cars still win !!
Problems are still unsolved !!
Renshi Luo (DCSC, TUD) 6 / 21
New approach for personal mobility
Renshi Luo (DCSC, TUD) 7 / 21
New approach for personal mobility
Cybercars
• automated road vehicles
Renshi Luo (DCSC, TUD) 7 / 21
New approach for personal mobility
Cybercars
• automated road vehicles
• provide on-demand and door-to-door service
Renshi Luo (DCSC, TUD) 7 / 21
New approach for personal mobility
Cybercars
• automated road vehicles
• provide on-demand and door-to-door service
• mostly small-sized
Renshi Luo (DCSC, TUD) 7 / 21
New approach for personal mobility
Cybercars
• automated road vehicles
• provide on-demand and door-to-door service
• mostly small-sized
• energy efficient
Renshi Luo (DCSC, TUD) 7 / 21
Cybernetic transportation system (CTS)
• promising solution to urban transportation challenges
Renshi Luo (DCSC, TUD) 9 / 21
Cybernetic transportation system (CTS)
• promising solution to urban transportation challenges
• formed by a fleet of cybercars
Renshi Luo (DCSC, TUD) 9 / 21
Cybernetic transportation system (CTS)
• promising solution to urban transportation challenges
• formed by a fleet of cybercars
• high flexibility amd reactivity
Renshi Luo (DCSC, TUD) 9 / 21
Cybernetic transportation system (CTS)
• promising solution to urban transportation challenges
• formed by a fleet of cybercars
• high flexibility amd reactivity
• emerged in Europe and first used at Schipol airport
Renshi Luo (DCSC, TUD) 9 / 21
Cybernetic transportation system (CTS)
• promising solution to urban transportation challenges
• formed by a fleet of cybercars
• high flexibility amd reactivity
• emerged in Europe and first used at Schipol airport
• projects including CyberCars, CyberMove and CyberCars-2,CityMobil
Renshi Luo (DCSC, TUD) 9 / 21
Opportunity and Challenge
Renshi Luo (DCSC, TUD) 10 / 21
Opportunity and Challenge
Opportunity
automated driving technology has been well developed
Renshi Luo (DCSC, TUD) 10 / 21
Opportunity and Challenge
Opportunity
automated driving technology has been well developed
Challenge
lack of efficient cooperation strategy
Renshi Luo (DCSC, TUD) 10 / 21
Cooperation of cybercars
Renshi Luo (DCSC, TUD) 11 / 21
Cooperation of cybercars
Cooperation
• necessary for optimal performance of CTS
Renshi Luo (DCSC, TUD) 11 / 21
Cooperation of cybercars
Cooperation
• necessary for optimal performance of CTS
variaty
• collision avoidance
• platoon merge and split
• dynamic routing
Renshi Luo (DCSC, TUD) 11 / 21
Cooperation of cybercars
Renshi Luo (DCSC, TUD) 12 / 21
Cooperation of cybercars
Multi-agent system
Renshi Luo (DCSC, TUD) 12 / 21
Cooperation of cybercars
Multi-agent system
• moving decision-making agents
Renshi Luo (DCSC, TUD) 12 / 21
Cooperation of cybercars
Multi-agent system
• moving decision-making agents
• extensive on-board processing and communication capabilities
Renshi Luo (DCSC, TUD) 12 / 21
Cooperation of cybercars
Multi-agent system
• moving decision-making agents
• extensive on-board processing and communication capabilities
• abundant information of environment
Renshi Luo (DCSC, TUD) 12 / 21
Cooperation of cybercars
Multi-agent system
• moving decision-making agents
• extensive on-board processing and communication capabilities
• abundant information of environment
Abstract object of our research !!
Renshi Luo (DCSC, TUD) 12 / 21
Fleet control problem of cybercars
Renshi Luo (DCSC, TUD) 13 / 21
Fleet control problem of cybercars
Existing reseach
Renshi Luo (DCSC, TUD) 13 / 21
Fleet control problem of cybercars
Existing reseach
• from conceptual point of view
Renshi Luo (DCSC, TUD) 13 / 21
Fleet control problem of cybercars
Existing reseach
• from conceptual point of view
• conceptual description of control system architecture andfunctions
Renshi Luo (DCSC, TUD) 13 / 21
Fleet control problem of cybercars
Existing reseach
• from conceptual point of view
• conceptual description of control system architecture andfunctions
Our research
Renshi Luo (DCSC, TUD) 13 / 21
Fleet control problem of cybercars
Existing reseach
• from conceptual point of view
• conceptual description of control system architecture andfunctions
Our research
• explore specific case, i.e., dynamic routing problem
Renshi Luo (DCSC, TUD) 13 / 21
Problem description
Renshi Luo (DCSC, TUD) 14 / 21
Problem description
Set up
Renshi Luo (DCSC, TUD) 14 / 21
Problem description
Set up
• dedicated roads network
Renshi Luo (DCSC, TUD) 14 / 21
Problem description
Set up
• dedicated roads network
• cybercars are free to make its desired route choices
Renshi Luo (DCSC, TUD) 14 / 21
Problem description
Set up
• dedicated roads network
• cybercars are free to make its desired route choices
• traffic densities in all roads decide velocities
Renshi Luo (DCSC, TUD) 14 / 21
Problem description
Set up
• dedicated roads network
• cybercars are free to make its desired route choices
• traffic densities in all roads decide velocities
• energy consumption depend on velocities and variation of
velocities
Renshi Luo (DCSC, TUD) 14 / 21
Problem description
Renshi Luo (DCSC, TUD) 15 / 21
Problem description
Modeling
Renshi Luo (DCSC, TUD) 15 / 21
Problem description
Modeling
• discrete-time
Renshi Luo (DCSC, TUD) 15 / 21
Problem description
Modeling
• discrete-time
• discrete-event
Renshi Luo (DCSC, TUD) 15 / 21
Problem description
Modeling
• discrete-time
• discrete-event
Objectives
Renshi Luo (DCSC, TUD) 15 / 21
Problem description
Modeling
• discrete-time
• discrete-event
Objectives
• minimizing total time spent (TTS)
Renshi Luo (DCSC, TUD) 15 / 21
Problem description
Modeling
• discrete-time
• discrete-event
Objectives
• minimizing total time spent (TTS)
• minimizing total energy consumption (TEC)
Renshi Luo (DCSC, TUD) 15 / 21
Model predictive control (MPC)
*figure from Ph.D thesis of L.D. Baskar
Renshi Luo (DCSC, TUD) 16 / 21
Centralized MPC
*figure from Journal paper of J.M. Maestre
Renshi Luo (DCSC, TUD) 17 / 21
Distributed MPC
*figure from Journal paper of J.M. Maestre
Renshi Luo (DCSC, TUD) 18 / 21
Research guide
Renshi Luo (DCSC, TUD) 19 / 21
Research guide
Modeling
• discrete-time modeling & discrete-event modeling
Renshi Luo (DCSC, TUD) 19 / 21
Research guide
Modeling
• discrete-time modeling & discrete-event modeling
Centralized control
• desired solution of distributed control
Renshi Luo (DCSC, TUD) 19 / 21
Research guide
Modeling
• discrete-time modeling & discrete-event modeling
Centralized control
• desired solution of distributed control
Distributed Control
• decomposing the problem
Renshi Luo (DCSC, TUD) 19 / 21
Research guide
Modeling
• discrete-time modeling & discrete-event modeling
Centralized control
• desired solution of distributed control
Distributed Control
• decomposing the problem
• extending exsiting distributed control methods
Renshi Luo (DCSC, TUD) 19 / 21
Research guide
Modeling
• discrete-time modeling & discrete-event modeling
Centralized control
• desired solution of distributed control
Distributed Control
• decomposing the problem
• extending exsiting distributed control methods
• developing new distributed control methods
Renshi Luo (DCSC, TUD) 19 / 21
Review
Renshi Luo (DCSC, TUD) 20 / 21
Review
• motivation
Renshi Luo (DCSC, TUD) 20 / 21
Review
• motivation
• specific research problem
Renshi Luo (DCSC, TUD) 20 / 21
Review
• motivation
• specific research problem
• potential control scheme
Renshi Luo (DCSC, TUD) 20 / 21
Review
• motivation
• specific research problem
• potential control scheme
• research guide
Renshi Luo (DCSC, TUD) 20 / 21
Renshi Luo (DCSC, TUD) 21 / 21