+ All Categories
Home > Documents > 4/15/20051 Simultaneous Design and Analysis of Communications and Control for Collaborative Teams of...

4/15/20051 Simultaneous Design and Analysis of Communications and Control for Collaborative Teams of...

Date post: 21-Dec-2015
Category:
View: 213 times
Download: 0 times
Share this document with a friend
Popular Tags:
39
4/15/2005 1 Simultaneous Design and Analysis of Simultaneous Design and Analysis of Communications and Control for Collaborative Communications and Control for Collaborative Teams of Unmanned Aircraft Teams of Unmanned Aircraft PhD Comprehensive Presentation Addendum Slides April 15, 2005 Cory Dixon Cory Dixon Research & Engineering Center for Unmanned Vehicles Aerospace Engineering Sciences University of Colorado
Transcript

4/15/2005 1

Simultaneous Design and Analysis of Simultaneous Design and Analysis of Communications and Control for Communications and Control for

Collaborative Teams of Unmanned Collaborative Teams of Unmanned AircraftAircraft

PhD Comprehensive PresentationAddendum Slides

April 15, 2005

Cory DixonCory DixonResearch & Engineering Center for Unmanned Vehicles

Aerospace Engineering SciencesUniversity of Colorado

4/15/2005 2

“The Department’s funding for UAV development has risen from just above $3 billion in the 1990s to over $12 billion for 2004 through 2009.”

http://www.whitehouse.gov/omb/budget/fy2005/defense.html

Are your taxes done Are your taxes done yet?yet?

“President Bush's FY '06 budget request for DOD includes between $1.7 billion and $2 billion for UAVs …”

Dyke Weatherington, head of the Pentagon's UAV Task Force http://aimpoints.hq.af.mil/display.cfm?id=1033

4/15/2005 3

OverviewOverview

• Introduction• Related Work

– Mobile Ad Hoc Networks (MANETs)– Collaborative Control

• Joint Design of Communication and Control

• Research Application Problem• My Experience & Previous Research

– AUGNet– Leashing– PIGF

• Research Timeline & Summary

4/15/2005 4

UAV Missions at CUUAV Missions at CU

AUGNet Leashing Problem

Tornado Chasing Artic Climate & Sea-Ice

Wild-land Fire

4/15/2005 5

Intelligence, Surveillance, Intelligence, Surveillance, and Reconnaissance and Reconnaissance

MissionsMissions

• Small UAVs provide two fundamental capabilities– In-situ sensing

• Atmospheric• EOI

– Aerial communications• Relay nodes for other UAVs or ground units.• RF eavesdropping and source localization/detection

• ISR missions are Dull, Dirty, Dangerous– Dull: ISR missions can be long, and uneventful– Dirty: UAVs can be sent to contaminated areas– Dangerous: No risk of human life, possibly expendable

• Common research interests– UAV Teams

• Collaborative teams can accomplish complex, multi-tasked missions • Teams are heterogeneous, utilizing capabilities of specialized vehicles

– Networked Communications• Provides remote operator/scientist with timely, high-level data and control• Enables communication among teammates, as opposed to neighbors

4/15/2005 6

Team Control and Team Control and Network Network

CommunicationCommunication

UAV Team ControlUAV Team Control

• UAV Autopilots– Low-level control of aircraft

dynamics– Provides high level interface

to aircraft control

• Team & Task Assignment– Allocation of resources– Assignment of task to

individuals within the team

• Consensus Problem– Method of coming to

agreement on action

Mobile Ad-Hoc NetworksMobile Ad-Hoc Networks

• MAC (Media Access Control)– Channel access method– Decentralized protocols

suffer from hidden terminals

• Network Routing– Provide address mechanism

to users– Determines routes between

sources and destinations – Route discovery

• Proactive• On-Demand• Probabilistic

• Quality of Service (QoS)– Throughput– Packet deliver– End-to-end delays

When the communication and control systems are When the communication and control systems are independently designed, there is no guarantee that the independently designed, there is no guarantee that the

combined system is capable of achieving the desired combined system is capable of achieving the desired performance and capabilities provided by the individual performance and capabilities provided by the individual

systems. systems.

4/15/2005 7

Communication and Communication and Control are InterlinkedControl are Interlinked

Network Protocols Affect Network Protocols Affect UAV Control & UAV Control &

OperationsOperations• Network Topology

– Point-to-Point– Multi-Hop

• Quality of Service– Bandwidth– Time Delays– Dropped Packets

• Mixed Traffic and Loads– Real-Time (Local Control)– Best-effort (Data Downlink)

Control of UAV Affects Control of UAV Affects Network PerformanceNetwork Performance

• Mobility affects network topology– Routes are time varying– May not have a communication

link

• QoS is impacted by physical link– Distances between vehicles– Antenna Gains and Pattern

• Neighbor communication– Adds congestion to network– Small packets with large headers

Joint design and analysis of the control and Joint design and analysis of the control and communication systemscommunication systems is required to utilize the is required to utilize the

full potential of collaborative UAV teams.full potential of collaborative UAV teams.

Network Performance

Cooperative Control

Network Performance

Node Mobility

4/15/2005 8

OverviewOverview

• Introduction• Related Work

– Mobile Ad Hoc Networks (MANETs)– Collaborative Control

• Joint Design of Communication and Control

• Research Application• My Experience and Previous Work

– AUGNet– Leashing– PIGF

• Research Timeline & Summary

4/15/2005 9

Mobile Ad Hoc Mobile Ad Hoc NetworksNetworks

• T. Camp– Effects of mobility on several routing protocols– Demonstrates different types of mobility models affect protocols in

different ways, generally reducing performance

• Woo and Culler – CSMA/CA with adaptive rate control scheme– protocol is based on concast network

• X. Hong et. al– Multicast Landmark ad-hoc routing (M-LANMAR)– Adapted to coordinated motion scenarios (teams)

• Qin and Kunz – Study impact of a realistic physical on several routing protocols– proposed new signal power thresholds for route discovery for AODV

and DSR

• MANETS (Wireless Sensor Networks)– MAC: IEEE 802.11, TDMA, FDMA, CDMA, Dual-Tone– Routing: DSR, AODV, DSDV, IGF, GR, ZRP– QoS: IEEE 802.11e, PSFQ

Network Performance

Node Mobility

4/15/2005 10

UAV Team ControlUAV Team Control

• T. W. McLain & R. W. Beard– Team Coordination and Consensus

• Coordination Variable• Coordination functions

– Communication Constraints• Cooperative search with collision avoidance• Coordinated target assignment

• R. M. Murray et al.– Graph theoretic approach

• Show link from graph Laplacian (topology) and the consensus problem (information flow)

• They prove a separation principle for formation stability– Stability of information flow graph– Stability of individual vehicle control

– Introduce geometric connectivity robustness

• Optimal Control & Estimation over Unreliable Communication Links– Imer et al.: Finite Horizon Optimal Control over TCP & UDP– Pollini et al.: Robustness to communication failures in formation flight– B. Sinopoli et al.: Kalman filtering with intermittent observations– P. J. Seiler: String stability on infinite vehicle platoons

Network Performance

Cooperative Control

4/15/2005 11

Research MotivationResearch Motivation

Combining ad hoc wireless networks with control of unmanned aircraft teams is not well understood at this time and there are two fundamental problems with current systems due to the fact that the communication and control systems are designed independently.

Network Performance

Cooperative Control

Network Topology

Node Mobility

– Communication systems that are currently used are not well suited for the mixed traffic types required by UAV teams and the highly dynamic network topology of UAV teams.

– Collaborative team controllers only consider limited, possibly incorrect, information about the underlying network communications and simply ignore the fact that the motion of the vehicle directly affects the performance of the network, and thus the performance of the control system.

Need to close the loop!Need to close the loop!

Network Performance

Cooperative Control

Network Topology

Node Mobility

4/15/2005 12

OverviewOverview

• Introduction• Related Work

– Mobile Ad Hoc Networks (MANETs)– Collaborative Control

• Joint Design of Communication and Control

• Research Application Problem• My Experience & Previous Research

– AUGNet– Leashing– PIGF

• Research Timeline & Summary

4/15/2005 13

Research ThesisResearch Thesis

• Communication should be developed specifically for UAV team control in ISR missions– Provide balanced traffic loading

• Hard real-time requirements from control system• Soft real-time and bandwidth guarantees for video and VOIP• Low priority data such as atmospheric measurements and UAV health & status

– Adaptable and robust to UAV mobility• Mobility is expected and possibly predictable

– Takes advantage of UAV team control• UAV Team hierarchy should be reflected in network routing• Knowledge of neighbor position is fundamental to network and control system

• Control system should consider communication performance as a controlled objective– Network connectedness is generally required

• Provides situational awareness from all UAVs to control center• Enables information sharing among all team members• Extended to include ground network connectedness in addition to team

– Network performance can be improved by UAV motion• Proper UAV dispersion will provide reliable link qualities and network coverage• UAV can find remote networks, and act as a data mule

Network Performance

Cooperative Control

Network Topology

Node Mobility

Network Performance

Cooperative Control

Network Topology

Node Mobility

Joint design and analysis of the control and Joint design and analysis of the control and communication systemscommunication systems is required to utilize is required to utilize the full potential of collaborative UAV teams.the full potential of collaborative UAV teams.

Example: In UAV team control, knowledge of neighbor positions is so fundamental that it should be considered at the lowest level of the communication system.

4/15/2005 14

Co-Design Co-Design FrameworkFramework

A proposed hierarchical framework in A proposed hierarchical framework in which network and control design can be which network and control design can be

considered simultaneously.considered simultaneously.

TeamControl

NetworkProtocols

Co-Design

MissionPerfomanceSpecifications

Co-SimulationSystem Integration

PerformanceCertification

yes

no

AdjustmentRules

PerformanceAchieved

4/15/2005 15

Research Goals & Research Goals & ContributionsContributions

• Develop theory for using mobility control to affect network performance– Position based control is not enough to guarantee a communications

link– Utilize communication-based information for control

• Signal-to-Noise Ratio (SNR)• Neighbor connectedness

– Balance communication performance with other mission objectives– Development of theory

• Co-design and analysis of communications and UAV team control– Introduction of a hierarchical framework for co-design– Development of simulation environment for co-analysis

• Development and testing of a jointly designed system– Develop MANET protocol and UAV team control scheme– Test on a physical system in addition to software simulation

4/15/2005 16

Related WorkRelated WorkGoldenberg et al.Goldenberg et al.

• Introduce a mobility control scheme for network performance– Decentralized control scheme

• Based solely on neighbor positions• Shown communication energy can be reduced by utilizing mobility control

– Introduce connectedness constraints on mobility• Constrained Mobility• Unconstrained Mobility

– Study energy performance on• Single unicast flow• Multiple unicast flows• Many-to-one concast flows

• Research Differences– Based on GPS position– Treats nodes as slow point masses (speed < 0.1 m/s)– Do not analyze network protocols, only use GR

“The energy optimal positions of relay nodes must lie entirely on the line between the source and the destination.”

4/15/2005 17

Related WorkRelated WorkM. Gerla and X. HongM. Gerla and X. Hong

• “Internet-in-the-sky”– Multimedia Intelligent Network of Unattended Mobile Agents (Minuteman)

• Landmark ad-hoc routing (LANMAR)• Mobile Backbone Network (MBN)• Distributed Information Database

– Adapted LANMAR • Exploited coordinated movements, e.g. teams• Included a multicast framework (M-LANMAR)• Scalable QoS with backpressure routing

• Research Differences– Do not control mobility– Designed independent of vehicle– Only present routing protocol

4/15/2005 18

Related WorkRelated WorkX. Liu and A. J. GoldsmithX. Liu and A. J. Goldsmith

• X. Liu’s Research: Joint Design of Control and Communications

• Previous Work– Kalman filter in presence of packet losses– String stability with communication delays– MAC & Link Analysis on Control

• Proposed a cross-layer framework to jointly design all the layers of the network to deliver the best end-to-end control performance

• Research Differences– No node mobility– Small static networks

=> TDMA is reasonable– Network control is application

4/15/2005 19

OverviewOverview

• Introduction• Related Work

– Mobile Ad Hoc Networks (MANETs)– Collaborative Control

• Joint Design of Communication and Control

• Research Application Problem• My Experience & Previous Research

– AUGNet– Leashing– PIGF

• Research Timeline & Summary

4/15/2005 20

Research Application:Research Application:Wildland FirefightingWildland Firefighting

• Situational awareness is a matter of life and death– UAVs can provide reliable communication links to ground crew– In-situ atmospheric measurements provide microscale meteorology– Aerial imagery provides an unmatched level of awareness as compared to any other sensor

• Typical deployment– Small / Micro Aerial Vehicles

• 1-10 UAVs for Type II Fires (county resources)• 10-50 UAVs for Type I Fires (Federal & State resources)

– Cary a variety of sensors• Atmospheric• EOI Sensors

– Ground Units• Control center• Mobile firefighting teams• Ground wireless sensor networks

• Mission Goals– Provide guaranteed communication from control center to all ground crew– Collect imagery for immediate downlink to control center– Combine microscale atmospheric measurements to generate a local weather map– Deploy ground sensors and gather information from them

• What improvements / capabilities will my system provide– Team performance will not be affected by video/VOIP data streams– Non-positional control will provide robust communication links to ground units– Communication chains can be established to enable long-range sensing

4/15/2005 21

Application Application DemonstrationDemonstration

• AUGNet testbed – Emulate Wildland Fire scenario on Table Mountain

• A Four plane Team– 2 equipped with cameras– All have atmospheric and communication equipment

• Single control center • Multiple ground networks

– Mobile crew– Static ground sensors

– Demonstrate UAV team control • Team and task assignment• Control based on communications => leashing

– Collect network performance in a physical environment• Utilize AUGNet tools to collect, monitor and store data• Show that the designed system performs better than was obtained by AUGNet

• Distributed Macro-sensor (DMS) Testbed– Show system is adaptable

• Can be applied to unmanned ground vehicles• Can scale by demonstrating on 20-50 vehicles

– Only if time allows as not primary thrust

Table Mountain National Radio Quiet Zone

4/15/2005 22

OverviewOverview

• Introduction• Related Work

– Mobile Ad Hoc Networks (MANETs)– Collaborative Control

• Joint Design of Communication and Control

• Research Application Problem• My Experience & Previous Research

– AUGNet– Leashing– PIGF

• Research Timeline & Summary

4/15/2005 23

AUGNet:AUGNet:AAd Hoc d Hoc UUAV AV GGround round

NetNetworkwork

• ObjectiveStudy ad hoc network behavior on a full-scale hardware test bed with mobile ground and air-vehicle nodes.

• Full-Scale Test Bed– Fixed nodes– Mobile ground nodes– UAV mounted nodes

• COTS Components– Single board computer– Linux OS– 802.11b PC Card– GPS

UAV Nodes Mobile Nodes

Meshed Radio Network

Fixed Site 1

Fixed Site 2

Test Bed Gateway and Test Range IP Router

University of Colorado

Monitor Server

Remote Monitor Internet

Table Mountain Field Site

Range Network

4/15/2005 24

AUGNet AresAUGNet Ares

5-hp Engine

Fuel Tank

Payload Bay with MNR

4/15/2005 25

AUGNet: AUGNet: Table Mountain Field SiteTable Mountain Field Site

4/15/2005 26

AUGNet 3-Plane TestAUGNet 3-Plane Test

4/15/2005 27

OverviewOverview

• Introduction• Related Work

– Mobile Ad Hoc Networks (MANETs)– Collaborative Control

• Joint Design of Communication and Control

• Research Application Problem• My Experience & Previous Research

– AUGNet– Leashing– PIGF

• Research Timeline & Summary

4/15/2005 28

The leashing problem is to provide a communication The leashing problem is to provide a communication link between two disconnected nodes (networks) using link between two disconnected nodes (networks) using

only information available from the established only information available from the established communication links.communication links.

Leashing ProblemLeashing Problem

Leashed chain for long-range sensingLeashed chain for long-range sensing

4/15/2005 29

Leashing ControllersLeashing Controllers

Use the signal-to-noise ratio (SNR) to determine control input.Use the signal-to-noise ratio (SNR) to determine control input.

Constant SNR Orbit Maximize & Equalize SNR

diIiiDiP

iiiiii

SSKeeKeKiuiu

rrN

KrSrSSe

1

10

01

1

11

ikin

k i

nk i

i GFamSktg

ktgG

1

1

Turning Rate Control Orbit Center Point Control

4/15/2005 30

Maximum-Equal SNRMaximum-Equal SNR

50 100 150 200 250 300 350 400

2

4

6

8

10

12

14

16

Time [s]

SN

R

Node 1Node 2Node 3Node 4

4/15/2005 31

OverviewOverview

• Introduction• Related Work

– Mobile Ad Hoc Networks (MANETs)– Collaborative Control

• Joint Design of Communication and Control

• Research Application Problem• My Experience & Previous Research

– AUGNet– Leashing– PIGF

• Research Timeline & Summary

4/15/2005 32

Prioritized Implicit Prioritized Implicit Geographic Geographic ForwardingForwarding

PIGF: A Real-Time Ad-Hoc Network Protocol for Micro-Air Vehicle Swarms

• Geographic Forwarding– UAVs already carry GPS – UAVs need to know neighbor’s positions for team collaboration and

collision avoidance

• Minimum Resources Available– Computational power– Power / Range

• Real-Time– End-to-End deadline– Single hop deadline

• Based on existing protocols– IGF– SPEED– RAP

• Open RTS

• CTS

• Data

• ACK

PrioritySource IDSourceDestination Location

Source ID

CSumDataSequence NumberSource IDDestination ID

Sequence NumberSource ID

7ACK

> 9DATA

5CTS

30ORTS

Size (Bytes)

Packet Type

7ACK

> 9DATA

5CTS

30ORTS

Size (Bytes)

Packet Type

4/15/2005 33

Implicit Geographic Implicit Geographic ForwardingForwarding

• IGF– MAC Layer routing– On demand routing without routing tables– Based on geographic position

• Open RTS (ORTS)– Candidate nodes compete to participate– Contains geographic destination of message

• CTS Response Time– Metrics

• Increased distance toward destination• Energy Remaining

– Can include any other measurable, local metric

RED

ED

WWW

randERCRIDTDCMSIFSERIDTDCTS

()1),(

sd

ORTSCTS

DATA

4/15/2005 34

SPEED & RAPSPEED & RAP

• SPEED provides real-time scheduling between nodes. It decides the candidate node to forward the packet to.

• RAP provides real-time scheduling within a single node. It decides the candidate packet to send out.

• They talk about real-time in different spaces:between nodes vs. within a node.

A C

ELec

Lac

ae

ecaccae T

LLS

p

acp T

LS

SPEED RAP

4/15/2005 35

OverviewOverview

• Introduction• Related Work

– Mobile Ad Hoc Networks (MANETs)– Collaborative Control

• Joint Design of Communication and Control

• Research Application Problem• My Experience & Previous Research

– AUGNet– Leashing– PIGF

• Research Timeline & Summary

4/15/2005 36

Future Work & MilestonesFuture Work & Milestones

• Refine Metrics– Efficiency – Robustness– Adaptability

• Integrate Comm & Control Simulation– Adapt current Matlab simulation to

cooperative control– Integrate and simulate PIGF in NS2– Link Matlab & NS2 Simulations to enable

co-analysis and design

• Leashing– Use communication as control primitive– Formulate team control problem

• Communication Impact on Control– How does communication affect control?

• Control Impact on Communication Network – How does UAV motion affect comm network?– Utilize graph theoretical tools

• Field Deployment & Tests– Implement comm and control on AUGNet

platform– Deploy on Table Mountain test site, using

AUGNet infrastructure to collect data

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Jan Feb Mar Apr May

Defense Graduation

2005

Control CommComm Control Field Deployment and TestsCommunications

Integrate Communication and Control Simulation

LeashingRefine Metrics & Thesis

Dissertation2007

2006Analysis Control Simulation

4/15/2005 37

SummarySummary

• Research Goals and Contributions– Develop an understanding of using mobility control to affect network performance

• Utilize communication-based information for control • Balance communication performance with other mission objectives

– Co-design and analysis of communications and UAV team control• Introduction of a hierarchical framework • Development of simulation environment for co-analysis• Development of theory to close the performance loop

• Experimental demonstration– Utilize AUGNet testbed for wildland fire scenario– Distributed Macro-sensor (DMS) Testbed to show adaptability and scalability

• Initial Experience and Research– Aerosonde– AUGNet– PIGF– Leashing

Simultaneous design of the control and communication Simultaneous design of the control and communication systemssystems is a significant research problem and is required to is a significant research problem and is required to realize the maximum performance and capabilities of UAV realize the maximum performance and capabilities of UAV

teams.teams.

Network Performance

Cooperative Control

Network Topology

Node Mobility

Network Performance

Cooperative Control

Network Topology

Node Mobility

4/15/2005 38

Questions and Comments are Welcomed

Thanks for [email protected]

4/15/2005 39

• Mixed traffic communication with• Multi-objective controller with

communication based input


Recommended