NetworkCentric Organization of MultiVehicle Systems

1
Network-Centric Organization of Multi-Vehicle
Systems

• Raja Sengupta
• Assistant Professor, Systems, CEE, UC Berkeley

2
Joint Work with.

• Dan Coatta
• Susan Dickey, PATH
• Tim McGee
• Mark Godwin
• ZuWhan Kim, PATH
• David Nelson, PATH
• Sivakumar Rathinam
• Allison Ryan
• Stephen Spry
• Xiao Xiao
• Marco Zennaro

• Eric Frew, Univ. of Colorado, Boulder
• Anouck Girard, Univ of Columbia
• Adam Howell, Lockheed Martin
• Aram Sogheikian
• Professor Pravin Varaiya
• Professor J. Karl Hedrick

3
UAV Network
4
UAV Network

• Objective
• Current 3 operators one unmanned plane
• One operator ? many UAVs
• Methodology
• Commander specifies what is to be done (missions)
• June 2005 Missions Visit location, Patrol border
• Autonomy System works out how it is to be done
• Mission decomposition into tasks
• A task is a unit of work for one UAV
• Allocation of tasks to UAVs
• Task execution Guidance, navigation, control

5
UAV Network

• Disturbances
• UAVs communicate over Wi-Fi ? channels come and
  go
• UAVs come and go
• Design objectives
• Robustness Event as platforms or channels come
  and go every task is eventually done by at least
  one UAV
• Should hold with weak assumptions
• Efficiency Completion time, throughput, jitter,
  fairness
• Should improve with stronger assumptions
• Network-centric organization
• Combination of ideas from P2P networks and
  control to achieve robustness and efficiency

6
Mission Control
7
Information Structure for Task Allocation

• Each agent computes and broadcasts the state of
  the mission
• Computation is based on the state received from
  the other agents
• The state of the mission

1
2
3
4
8
Current Collaborative Architecture

• Mission statements are decomposed into tasks and
  relayed from the ground to all aircraft.
• Each plane without a task picks the closest
  available task for itself. Each plane allocates
  tasks only for itself.
• Conflicts in task allocation are resolved using
  Euclidean distance.
• Each agent can write its own component of the
  mission state
• Its own position and task it is doing
• It can add tasks to the to-do list
• Conflicts in information are resolved by GPS
  timestamps
• Example If the position of blah in message A is
  more recent than that in message B then I will
  write the position in message A for blah into my
  broadcast

9
Commander View
10
Generalization Vision Based Following of Locally
Linear Structures(Closed Loop on the California
Aqueduct, June 2005)
11
Flying at a target
12
System Level Architecture
Mission Control
Ground-to-Air Communications
Air-to-Air Communications (802.11b)
Collaboration
Collaboration
Switchboard
Switchboard
UAV Piccolo Autopilot
UAV Piccolo Autopilot
Aircraft 1
Aircraft 2
13
Aircraft Level Architecture
Payload Responsible for Relaying Commands between
the PC-104 and Mission Control
Piccolo Responsible for Relaying Commands between
the PC-104 and Aircraft Avionics
Database Permits inter-process communication
Vision Processing Frame Grabbing Capabilities
Orinoco Inter-vehicular communication protocol
Switchboard Task Allocation, Conflict Resolution,
and Controller Switching
Vision Control For Turn-Rate Based Path Following
Waypoint Control For Single-Point Visits
Orbit Control For Closed-Loop Multi-Point Paths
14
Current UAV Platform Configuration

• Wing-Mounted Camera allowing for vision-based
  control, surveillance, and obstacle avoidance
• Ground-to-Air UHF Antenna for ground operator
  interface
• GPS Antenna for navigation
• 802.11b Antenna for A-2-A comm.
• Payload Tray for on-board computations and
  devices
• Payload Switch Access Door for enabling /
  disabling on-board devices

15
Current Payload Configuration

• Off-the-shelf PC-104 with custom Vibration
  Isolation
• Orinoco 802.11b Card and Amplifier for A-2-A
  comm.
• Analog Video Transmitter for surveillance
  purposes
• Printed Circuit Board for Power and Signal
  Distribution among devices.
• Umbilical Cord Mass Disconnect for single point
  attachment of electronics to aircraft.
• Keyboard, Mouse, Monitor Mass Disconnect for
  access to PC-104 through trap door while on the
  ground.

16
Airframe Specifications- Bat IV

• Weights and payload
• 45 lbs dry airframe
• 12 lbs fuel (2 gal)
• 25 lbs payload
• Flight envelope
• Vmin 35 kts (40mph, 18 m/s)
• Vmax 70 kts (80mph, 36 m/s)
• Turn rate lt 15 deg/sec
• (_at_Vmin,30 deg bank)
• Endurance
• Nominal 8 hrs
• Batt. Only 2-3 hrs
• (with payload shut down)

17
Salient Feature

• Network-centric relies on Information-centric
• Coordination arises from
• the information structure
• the decision making rule in each agent
• Agents do not command agents
• Our car network is organized similarly

18
Car Networks

• Similar Architecture
• P2P, estimation, broadcast communication

19
Wireless and Embedded Computing ? Enabling New
Vehicle Roles

• Vehicle as a producer of geo-referenced data
  about its environment
• Pavement condition
• Probe data for traffic management
• Weather data
• Physiological condition of passengers, .
• Vehicle Vehicle, Vehicle Roadway as
  collaborators
• Cooperative Active Safety
• Forward Collision Warning, Blind Spot Warning,
  Intersection Collision Warning.
• In-Vehicle Advisories
• Ice on bridge, Congestion ahead,.
• Vehicle as Information Gateway (Telematics)
• Internet access, infotainment, dynamic route
  guidance,

20
Driving the ConvergenceGovernment, Industry,
Academia

• ACM created Vehicular Ad-hoc Networks Workshop
• IEEE created V2VCOM
• Federal Communications Commission created DSRC
• The record in this proceeding overwhelmingly
  supports the allocation of spectrum for DSRC
  based ITS applications to increase traveler
  safety, reduce fuel consumption and pollution,
  and continue to advance the nations economy.
• FCC Report and Order, October 22, 1999, FCC
  99-305
• Amendment with licensing rules in December 2003
• DSRC Standards
• ASTM E17.51, IEEE 802.11p
• http//grouper.ieee.org/groups/scc32/dsrc/
• Automotive companies created Vehicle Safety
  Communications Consortium (VSCC)
• Final Report Submitted January 2005
• USDOT/CAMP have created Cooperative Intersection
  Collision Avoidance (CICAS) Consortium
• http//www.its.dot.gov/cicas/cicas_workshop.htm

21
USDOT Vehicle Infrastructure Integration
Initiative

• http//www.itsa.org/vii.html
• The VII Initiative is a cooperative effort
  between Federal and state departments of
  transportation (DOTs) and vehicle manufacturers
  to evaluate the technical, economic, and
  social/political feasibility of deploying a
  communications system to be used primarily for
  improving the safety and efficiency of the
  nation's road transportation system.

22
(No Transcript)
23
(No Transcript)
24
Led to Research on Active Safety Systems
targeting the Different Crash Types

• Initial Approach
• Sensor based
• Multiple sensors for 360 coverage
• Cost remains significant
• More Recently
• Cooperative Active Safety
• Wireless equipped vehicles communication GPS
  coordinates
• Cheap

Blind-spot radar
Forward Collision Radar
Mid-Range
Backing radar
Mid-Range
25
(No Transcript)
26
More Futuristic Fully automated drivingThe
Wireless Token Ring Protocol
27
Network-centric Organization of Homogeneous
Systems

• The UAV and Car networks are Homogeneous
• Identical agents
• Network-centric organization by combining P2P
  with estimation, control
• Tolerates channels and agents appearing and
  dissapearing
• Keeps some properties invariant to changes of
  structure
• Cars Threatening neighbors get tracked with high
  probability (High?)
• UAVs Every task is usually eventually done
  (usually?)
• New design tools required

28
New Design Tools Tight real-time control
requires new MAC protocols
T
Data
PS (C) B NS (C) A
C
T
Data
T
Data
B
PS (B) A NS (B) C
A
PS (A) C NS (A) B
Ergen M., Lee D., Sengupta R., Varaiya P.
Wireless Token Ring Protocol. IEEE transactions
on Vehicular Technology, vol.53, no.6, Nov. 2004,
pp.1863-81. Ergen M., Lee, D. Datta R., Ko J.,
Puri A., Sengupta R., Varaiya P. Comparison of
Wireless Token Ring Protocol with IEEE 802.11.
Journal of Internet Technology, Volume 4, no. 4,
February 2004.
29
Network performance and Control performance
Worse spacing control ?
Worse network performance?
30
Feasibility by Simulation Loss rate lt 0.01,
Channel Busy Time lt 50
31
Broadcasting at max rate with max range all the
time is not feasible with this network design

• 50 msec, 300 meter not feasible
• Feasible averages
• 200 byte
• 200 msec
• 150 meter

4 lanes, capacity, 150 m
4 lanes, capacity, 300 m
32
Range/Power need to adapt to traffic density

• Feasible Averages
• 200 byte
• 200 msec
• Range
• Max flow
• 150 m
• Jammed
• 50 m

4 lanes,jammed, 50 m
4 lanes, capacity, 150 m
33
Basic principles of efficient design not known

• Is there loss of performance by separation of
  source and channel coding?
• Multiple access design
• Each car wants to keep its tracking errors within
  the same bounds
• What is the right channel sharing protocol?

34
Multi-Vehicle Network Control May Need New Models
of Computation
Synchronous Model of Computation
Caspi P. Embedded Control From Asynchrony to
Synchrony and Back, EMSOFT 2001
35
The Synchronous Model of Computation Semantics
Cascade Composition
Berry 91
yf(u)
ug(y)
Feedback Composition Fixpoint Semantics
36
First result(Zennaro, Sengupta, EMSOFT 2005)

• We define the map ?1 from RA to FSTS traces as
  in
• Implementation given the map ? from RA to FSTS
  traces the implementation map ? has the following
  property for all FSTS s and RA r the following
  holds
• r ?(s) ? (r t ? s ?(t))
• That is to say the RA r that implements the
  FSTS has the same set of behaviors of s.

1 Benvenieste, Caillaud, Le Guernic
Compositionality in dataflow synchronous
languages specification and distributed code
generation, Information and Computation, col.
163, Nov 2000, Academic press
37
Second result Distribution and Modular
Compilation

• Monomorphism ? is a monomorphism between (FSTS,
  xSTS) and (RA, xRA). The following must hold for
  all FSTS s1 and s2 and RA r1 and r2

38
Network-Centric Organization for Heterogeneous
Systems

• Semantics unclear

39
Overview

• Objectives are the same but the system is made of
  different kinds of components
• Network-centric organization
• Certain types of components can come and go but
  the mission survives
• Challenges
• How to insert a new agent into the system?
• What connections does it need? What controllers?
• This now depends on the agent and the mission
• Approach Middleware (Jini)

40
What is Jini?

• Vehicles and tasks appear at different times
• They connect anyway
• Easy repair
• UAV fails task is picked up by another in the
  swarm
• Proxy implementation invisible to user
• Easy system integration

Registration
Service
Lookup manager
Multicast Network
RMI
Proxy
ServiceItem
Client
41
One Concrete Service NetworkMulti-vehicle
Search Mission Architecture
Mission control
Search
42
Multi-vehicle Search Mission Architecture
Mission control
Search
Sweeper
43
Multi-vehicle Search Mission Architecture
Mission control
Results
Search
Sweeper
44
Multi-vehicle Search Mission Architecture

• There is a mission control program running on a
  mission control computer that wants to search for
  a robot. It allocates vehicles to be used for the
  search. The search is executed by a program that
  accepts a specification of the robot and of the
  area to be searched. It coordinates a set of
  sweepers. To do this, it partitions the area to
  be searched into sectors. The vehicles allocated
  by mission control offer the sweep service. They
  are capable of sweeping a sector and reporting
  the result of the sweep back to the search
  program. The search program publishes the set of
  sectors to be searched. Each sweeper takes a
  sector from this set, sweeps it, and reports the
  result of the sweep back to the search program.
  The search program evaluates the results coming
  back in and changes the set of sectors to be
  searched if necessary.

45
Systems with Changing DimensionService Network
for Multi-Vehicle Search
Lookup Manager
Mission Control
Search Software
Host
Vehicle
Txn Manager
Search Server
Publisher
Sengupta CDC 2002
46
The Search Mission Screen Dump
47
Scalable Information ManagementDistributing the
Publisher Service

• Geographic Data Management Network

Sengupta AINS 2003
48
Scalable Information ManagementDistributing the
Publisher Service
User
49
Movie of Implementation

• 4 laptops over wireless
• One publisher per laptop
• Start with one publisher
• Three others come up
• Some die
• Data redistributes as publishers join and leave

50
Movie of Implementation
Total data made of many data objects

• 4 laptops over wireless
• One publisher per laptop
• Start with one publisher
• Three others come up
• Some die
• Data redistributes as publishers join and leave

51
Movie of Implementation

• 4 laptops over wireless
• One publisher per laptop
• Start with one publisher
• Three others come up
• Some die
• Data redistributes as publishers join and leave

52
Movie of Implementation

• 4 laptops over wireless
• One publisher per laptop
• Start with one publisher
• Three others come up
• Some die
• Data redistributes as publishers join and leave

53
Movie of Implementation

• 4 laptops over wireless
• One publisher per laptop
• Start with one publisher
• Three others come up
• Some die
• Data redistributes as publishers join and leave

54
Scalable Information ManagementDistributing the
Publisher Service
Movie of our Implementation 4 servers on 4
laptops over wireless
55
Data Consistency in the PublisherInconsistent
copies are detected whp
Wrong location copy 1
Data Location
Wrong location copy 2
56
Data Consistency in the PublisherDrift in a 2-D
Markov Process
57
Movie of Pursuit Evasion

• Vehicle simulated on laptops over wireless LAN

58
The SN Synthesis Theorem

• Theorem Let the set of concrete values be
  finite. Then there exists a synthesis procedure
  that is sound, complete, and decidable.

59
Scalable Information ManagementDistributed
Estimation over the GDMN
Theorem 2 If observations stop coming in at time
m and reach all the map-builders by time n,
then If communication between servers is faster
than the target then the Distributed computations
converge to the centralized ones.
Movie of our Implementation 4 servers on 4
laptops over wireless
Mahajan, CDC 2002
60
Summary Network-centric Organization of
Networked Multi-Vehicle Systems
Verified (stability,safety, fairness,optimality)
Design Syntax
Semantics
Execution on Sequential Machine Networks
compiled
approximate

• Homogeneous systems
• Heterogeneous systems

61
Summary Network-centric Organization of
Networked Multi-Vehicle Systems
Verified (stability,safety, fairness,optimality)
Design Syntax
Vector Trace
Sequential Machine
compiled
approximate
P. Varaiya. A question about hierarchical
systems. System Theory modeling, analysis and
control, Kluwer, 2000.
62
Summary Network-centric Organization of
Networked Multi-Vehicle Systems
Verified (stability,safety, fairness,optimality)
Design Syntax
Vector Trace
Sequential Machine
compiled
approximate
P. Varaiya. A question about hierarchical
systems. System Theory modeling, analysis and
control, Kluwer, 2000.
63
Summary Network-centric organization of
Multi-Vehicle Systems