Path Planning and Obstacle Avoidance Strategies for … asokan.pdf · Path Planning and Obstacle...

54
Path Planning and Obstacle Avoidance Strategies for Autonomous Underwater Vehicles Prof. T. Asokan Indian Institute of Technology Madras [email protected]

Transcript of Path Planning and Obstacle Avoidance Strategies for … asokan.pdf · Path Planning and Obstacle...

  • Path Planning and Obstacle Avoidance Strategies for

    Autonomous Underwater Vehicles

    Prof. T. AsokanIndian Institute of Technology Madras

    [email protected]

  • Areas of Research and development

    1. Autonomous Underwater Robots: Design, Modelling, and Control

    2. Dynamic station keeping control of ROV3. Multipoint Potential Field method for

    obstacle avoidance

    4. AUV-Manipulator Systems

    5. Surgical Robotics- Design and Development of Tele-surgical trainer robot (Filed 2 patents)

    6. Tremor compensation in surgical robotics7. Aerial Robotics: Design of Novel quadrotor-

    VOOPS (Filed 2 patents)8. Underwater mechatronic system for

    selective deployment (Filed 1 patent)9. Task allocation in multirobot systems

  • Product Design and Development

    1. Limb Immobilisation splint for trauma victims Patented; Licensed to and commercialised by M/s HLL Life care (Asokan, Rahul Ribeiro, Pulin, Darshan)(first product to be commercialized out of the Stanford-India Biodesign program)

    2. Combinational Scissor-Grasper for laparoscopic surgery(Asokan, Chinmay Deodhar)Patented and Licensed to M/s Intuitive Surgicals,

    USA

    3. Wound cleaning Robot for hospital application

    4. Non-Destructive Method to Identify Used Syringes (Patent filed in 2012; innovation award from M/s Intellectual ventures; JC Bose Patent award

    2013)

  • 5. A device for descaling the inner wall of a tank (Patent filed in 2008)

    6. Fetal Growth Monitor for Rural Application

    7. Proportional solenoid for space applications (ISRO)

  • Acknowledgements

    Some of the works presented are from the published works of research students. The contributions by the following research scholars/ organisations are thankfully acknowledged:

    Dr M Santhakumar

    Dr T Periasamy

    Dr S Saravanaklumar

    NSTL Vizag

  • Contents

    Introduction

    Underwater Robots

    Atonomous Underwater Vehicles

    AUV Design

    Design of Controllers

    Path Planning

    Obstacle Avoidance

    Experimental Methods

    Conclusions

  • UNDERWATER WORKING METHODS

  • 3/17/2016 8

    Why Underwater Robots ?

    Motivation

    US$300 B offshore industry Demand for pipeline survey and

    inspection New installation of underwater pipeline Periodic checking required of installed

    pipeline Task is performed under structured

    conditions Extended duration of surveys

    up to weeks for extensive pipelines boredom and fatigue

    Major Players:

    Institutes: University of Tokyo, MIT, Hawai University, WoodsholeOceanographic Institute, etc

    Industry:

    Alstom Schilling, Perry Tritech, TritechInternational etc..

  • Human Occupied Vehicle (HOV)

    ALVINWoods hole Oceanographic Institute

  • 3/17/2016 10

    Unmanned Robotic Vehicles

    OperationMode

    Task Complexit

    y

    Tele-operation Supervisory Autonomous

    ROV

    HybridURV

    AUV

    Hybrid URV canswitch to direct tele-operation mode forcomplex tasks

    Hybrid URV can switch to autonomous mode for simple tasks

    Comparison of Task Complexity

  • Remotely Operated Vehicle (ROV)Tethered Supervised Vehicle:

    The vehicle is connected to a mother ship by a cable through which communications, data transmissions and power supply are carried out.

    Restricted operating range

    Mother ship to follow the ROV

    Pilot needs to be trained well

    Operator fatigue affects mission results

    Tether management is the main challenge

    Operationally efficient compared to other class of vehicles

  • ROV Deployment and Applications

    Diver Observation Platform Inspection Pipeline Inspection Surveys Drilling Support Construction Support Debris Removal Platform Cleaning

    Sub-sea Installations

    Telecommunications Support Object Location and Recovery

  • Autonomous Underwater Vehicle (AUV)

    It is a robotic device that is driven through the water by apropulsion system, controlled and piloted by an onboardcomputer and maneuverable in three dimensions.

    It needs to be Pre-programmed

    Degree of human intervention will be a function of communication capability

    AUV will require fool-proof navigation, control and guidance systemson board to meet the mission accuracy requirements

    Transmission of data back to mother ship if on-board data storage with post mission retrieval does not meet the mission requirements

  • Autonomous Underwater Vehicle (AUV)

  • Geological Survey Earthquake detection and

    monitoring. Tsunami detection. Volcano monitoring. Salinity monitoring. Seabed and deep sea

    exploration. Oil and Gas Industry

    Seabed survey. Pipeline monitoring. Intervention. Iceberg reconnaissance.

    Marine Science Research Marine Biology Research Oceanography Studies

    Shipwreck Reconnaissance Fisheries

    Academic Research. Development. Teaching. Practical applications. Projects.

    Defense Reconnaissance. Monitoring. Detection. Surveillance.

    Customs Underwater power lines

    Line laying. Line monitoring

    Delivery Systems Telecommunications Repair and Rescue

    Applications of AUVs

  • Main Building blocks of AUV Control System

    Data loggingData logging

    management

    CommunicationsRF and Acoustic modem

    NavigationRate gyro,

    Compass,

    Depth or Pressure sensor,

    Doppler log,

    Echo sounder,

    GPS

    Mission ControlOcean Sensors

    Vehicle SafetyActuator Control

    Thrusters, rudder and

    stern planes

    Vehicle

    Guidance and

    Control

    Set

    points

    Reference

    Trajectory

    Vehicle

    Dynamics

    NavigationSystem

    support data

  • Control, Guidance, and Navigation

    Where am I going? How do I get there?

    Control: To ensure that the various performance parameters are achieved in the system: eg: Speed, Acceleration, etc.

    Guidance: To ensure that the commanded path is followed by the system

    Navigation: To inform the MCS about the present status of the vehicle wrt the commanded position as well as react to the environmental changes

  • 18

    Mathematical Model of the AUV

    where M = MRB+MA = Inertia matrixMRB = Rigid body Inertia MatrixMA = Added mass matrixD() = Damping matrix = Control inputs = Bf = FT + FCPFT = Propulsion forces and momentsFCP = Control plane forces and momentsC() = C()RB+ C()A = Coriolis and Centripetal Matrixg() = Restoring matrix (gravitational and buoyancy effects)

    = Linear and angular velocities w.r.t. body (moving) frame = [u v w p q r]T

    = Linear and angular displacements w.r.t. inertial (fixed) frame= [x y z ]T

    The AUV kinematics can be expressed as

    )g()D()C(M

    )(J where, J() is the kinematic transformation (Jacobian) matrix.

    f)(B)(g),(D),(C)(M

    The AUV dynamic model with respect to the earth fixed frame of reference becomes

  • )))))((1 g(D(C(M

  • 0

    5

    10

    15

    20

    25

    30

    -25

    -20

    -15

    -10

    -5

    0

    5

    -100

    0

    100

    200

    300

    400

    500

    600

    700

    800

    x in m

    3D plot

    y in m

    z in m

    Only one propulsion thruster activated, CB=[0 0 -7]mm,& CG=[0 0 0]mm and the vehicle net buoyancy effect is equal to+25kgf

    0 100 200 300 400 500 600 700 800 900 1000-100

    -80

    -60

    -40

    -20

    0

    20

    Time (t) in Sec

    Angula

    r D

    ispla

    cem

    ent

    in d

    eg

    Angular Displacement vs Time

    Roll

    Pitch

    Yaw

  • 21

    Closed-loop (feedback) Control

    Feedback

    Control system

    Environmental

    Disturbances

    AUV Dynamics AUV Kinematics

    Thruster

    Forces and

    moments

    JM-1

    g(.)

    C+D

    Thruster

    dynamics

    d

    d

    Rudder &

    stern plane

    Forces and

    moments

    Control planes

    dynamics

    GNC

    Sensor

    System

    Mission

    control

    d

    Sensor

    Noises

    +

    +

    +

    +

    +_+

    _

    +

    e

  • Vehicle Control System

  • 23

    Tracking Controller Design

    Model reference control structure

    Hybrid control structure

    Tracking control structure

    Trajectory tracking requires the design of control laws that guide the vehicle to

    track an inertial trajectory, i.e., a 3D path on which a time law is specified.

  • Guidance System

    Development of strategy and algorithm for obstacle avoidance and

    path planning

    Guidance Systems: Receive target related informationfrom the navigation and generate references for the vehicle

    control system so as the vehicle can move through a set of

    way points as per the given sequence.

    Time variant trajectory tracking or time invariant path

    following.

    Guidance systems include waypoint guidance, obstacle

    avoidance, minimum time navigation, fuel optimization and

    weather routing.

  • Way-point guidance by Improved LOS method

    Both next and the previous waypoints are

    considered.

    Auxiliary points are fixed on both sides of the

    waypoints.

    A distance threshold is fixed at both sides of the

    waypoint

    Vehicle calculates the heading correction only

    between the threshold points and waypoints.

    Heading correction is calculated when AUV is

    moving towards a waypoint as well as starts

    leaving the same waypoint.

    Desired heading= reference heading +

    correction heading.

    3/17/2016 26

    2D path following

    WP1

    WP3

    WP2o

    VP

    r

    Circle of acceptance

    d

    Methodology for 2D LOS

    method

  • Starting Point: (50,50)

    Way Point: (0,200),(300,350),(450,50),(300,200) and (100,150)

    Goal Point: (100,0)

    3/17/2016 27

  • Animation 1

    Animation 2

    ../Videos/los4.avi../Videos/LOS5.AVI.avi

  • Figure : Calculation of heading correction for smooth turn

    during course change at waypoints.

    Methodology for algorithm development

    crd

    - desired heading,- actual reference heading- heading correction

    d

    r

    c

    mLwhereLThreshold

    Lo

    5.4,*10

    ,*2

    ),( iii yxWP

    ),( AiAii yxA

    idA

    idP

    Aio),( BiBii yxB

    idB

    GP

    ),( )1()1(1 iii yxC

    ),( iii yxD

    ),( iii yxC

    ),( )1()1(1 iii yxD

    ),( 111 iii yxWP

    SP

    cr

    r

    rcVP

    idP CidP

    DidP

    3/17/2016 29

  • 30

    Distance between vehicle position and auxiliary point Ai:

    2

    vAi

    2

    vAi

    2

    vAiAi )zz()yy()xx(d

    o

    CiAiAi

    dPd

    Normalized difference between the auxiliarypoint Ai and waypoint from the vehicle position:

    Distance between vehicle position and current way-point

    : 2vi

    2

    vi

    2

    viP )zz()yy()xx(d ci

    Sign of orientation correction between Ci and WPi :

    )(f).dP(f).(

    )(f).dP(f).(

    AiaCiACscc

    AiaCiACscc

    [Bakaric et al. (2004)]

    Orientation correction between Ci and WPi:

    Desired orientation can be calculated as

    Map the spherical coordinate to 3D Cartesian coordinate

    )sgn(

    )sgn(

    1ririsci

    1ririsci

    ,ciridi

    ciridi

  • No

    Ye

    s

    No

    Ye

    s

    Ye

    s

    Ye

    sNo

    No

    DETERMINE: The reference heading

    and path angle in spherical coordinatesDEFINE: Distance threshold points at

    both sides of the current waypoint

    FIND: The sign of correction heading

    and path angle to be required for

    smooth turn

    COMPUTE: The correction heading

    and path angle to be required for

    smooth turn

    CALCULATE: The desired heading

    and path angle and map the spherical

    coordinates to Cartesian coordinates

    Is current vehicle

    position =

    waypoint1?

    Start

    Is number of

    waypoints is

    equal to zero?

    DEFINE: Auxiliary points at both

    sides of the current waypoint

    DEFINE: Distance and Angle factors

    by linear curve fitting

    Is current vehicle

    position = goal

    position?

    DEFINE: Aiming point distance,

    reference angles and instantaneous angles

    BRING: Vehicle position to coincide with

    the tangential line to current waypoint

    Is current vehicle

    position= first

    distance threshold

    point?

    Stop

    OBTAIN: Vehicle and Goal position,

    previous, current and next waypoints in

    3D Cartesian coordinates

    SELECT: The next waypoint

  • 3/17/2016 32

    0

    50

    100

    150

    0

    20

    40

    60

    80

    0

    2

    4

    Surge(x),[m]Sway(y),[m]

    Heav

    e(z

    ),[m

    ]VP

    GP

    WP

    Basic LOS

    C1

    D1

    A1

    B1

    Improved LOS

    0

    100

    200

    300

    400

    500

    0

    100

    200

    300

    400

    500

    0

    100

    200

    300

    400

    500

    Surge(x),[m]Sway(y),[m]

    Hea

    ve(

    z),[

    m]

    Basic LOS

    Improved LOS

    0

    100

    200

    300

    0

    50

    100

    150

    0

    10

    20

    30

    40

    50

    60

    Surge(x),[m]Sway(y),[m]

    Hea

    ve(

    z),[

    m]

    Basic LOS

    Improved LOS

    SP: (10,0,2), WP: (0,0,0) & (75,0,75) GP: (150,0,0)

    SP: (10,0,2), WP: (0,0,0) & (75,75,0) GP: (150,0,0)

    SP: (20,50,0), WP: (0,0,0),(75,75,75), (150,150,150), (250,250,250) & (300,300,300), GP:

    (500,500,500)

    Simulation results

    SP: (150-20,0), WP: (150,-20,0),(0,0,50), (350,0,50), (200,50,50) (350,100,50) & (10,100,50), GP:

    (150,120,0)

    0

    50

    100

    150

    -5

    0

    50

    20

    40

    60

    80

    Surge(x),[m]Sway(y),[m]

    Hea

    ve(

    z),[

    m]

    Basic LOS

    Improved LOS

  • 0 20 40 60 80 100

    0

    5

    10

    15

    20

    25

    30

    35

    40

    45

    50

    Surge(x),[m]

    Sw

    ay(y

    ),[m

    ]

    Basic LOS method

    Smoothing spline for basic LOS

    Bakaric method

    Smoothing spline for bakaric method

    Improved LOS

    Smoothing spline for Improved d LOS method

  • Multipoint Potential Field Method for Obstacle Avoidance

  • Multipoint potential field method for Obstacle

    avoidance in 3D space

    factorscalingpositiveais

    intpogoalandvehiclethearound

    intpoqbetweencetandistheisqqdth

    iiaa

    2

    aatt iid

    2

    1)q(U

    planeverticalinangleturningMaximum

    planehorizontalinangleturningMaximum

    m

    m

    aq(b)

    v

    v

    r

    z y

    x

    Goal

    1q

    42q

    vA

    41q

    49q

    mhv

    mhv

    41ad

    z

    y

    x

    Goal

    h

    h

    Z

    YX Reference

    Frame

    Body Fixed Frame

    (a)

    [Ge and Cui

    (2000)]

    Attractive potential Uatt at qi points:

    3/17/2016 SSK,IITM 35

  • r

    r

    R

    zy

    Goal

    vA

    41q

    90

    90

    hhm

    hhm

    x

    3

    7

    1

    4

    2

    6

    5

    1

    5

    34153od

    4113od

    to

    to

    2

    a

    2

    to

    obs

    ddif,0

    ddif,dd

    1

    d

    1

    2

    1

    )q(U

    j,i

    j,ii

    j,i

    j,i

    factorscalingpositiveais

    thresholdluenceinfcetandistheisd,obstacletheon

    intpopandintpoqbetweencetandistheis|pq|d

    t

    th

    j

    th

    ijio

    sensorofwidthbeamvertical

    sensorofwidthbeamhorizontal

    bm

    bm

    [Ge et al.(2000)]

    Repulsive potential Uobs at qi due to the obstacle

    point pj:

    3/17/2016 36

    sA

    Calculating radius of

    obstacle

    Read Input: Sensor data

    Generate multiple points on the obstacle

    Calculate repulsive potential for the obstacles

  • Total repulsive potential Urep at qi due to

    the obstacle:

    Total potential Utot at each point around

    the vehicle:

    The above steps can be written in a single

    equation as:

    Minimum potential:

    Command the vehicle to the minimum

    potential position

    Local minima problem

    3/17/2016 37

    )q(U)q(U

    K

    1j

    obsrep j,ii

    R

    1m

    repatttot )q(U)q(U)q(U m,iii

    to

    to

    2

    a

    2

    toK

    1j

    R

    1m

    atttot

    ddif,0

    ddif,dd

    1

    d

    1

    2

    1

    qU)q(U

    m,j,i

    m,j,ii

    m,j,i

    ii

    )Umin(U tottotmin

    rvCZ and,R Critical (local minima) zone in

    horizontal plane is defined by

    where,

    ,and,R vrCZ

    Critical zone in vertical plane is defined by:

    fauvobsCZ dRRR

    Rcz- range of critical zone, Robs- radiusof obstacle, Rauv- radius of AUV, df-distance factor

    Divide each zone into 3 sectors

    Check for the presence of obstacle in the

    sectors

    Declare local minima, if exists

    Activate strategy to avoid

  • Yes

    No

    Yes

    No

    Stop

    Start

    GIVEN: Starting and Goal

    positions

    DISCRETIZE: The region on a

    hemisphere of radius r meter

    around the vehicle into equiangular

    points

    DEFINE: Critical Zone (CZ) for

    checking local minima

    CALCULATE: Attraction potential

    (Uatt)

    CALCULATE: Repulsion potential

    by discretizing obstacle into K2

    points (Uobs) for each obstacle

    FIND: The complete Repulsion

    potential (Urep) for the obstacles

    CALCULATE: Total potential

    (Utot) for each point around AUV

    DETERMINE: Minimum total

    potential (Umintot)

    COMMAND: Move the vehicle

    towards the minimum potential

    point (Umintot)

    GET: Obstacles data from sensor

    DECLARE: local minimum exists and

    activate hovering thruster

    Does Local

    minima Exist?

    Is current

    position= goal

    position

  • Simulation studies

    Configuration: Flat-fish type

    Design speed: 4 knots (max)

    Buoyancy: Positive

    3/17/2016 39

    Both static and dynamic obstacles are assumed and they are in spherical shape of various

    sizes. The vehicle is reduced in size to a single point, and the obstacles are enlarged by the

    vehicles radius.

    Two forward looking sonar sensors are considered to sense the obstacle. One sensor with a

    range of 40m, horizontal & vertical beam widths of 1200 and 300 is fixed horizontally.

    The other sensor with a range of 40m, horizontal & vertical beam widths of 300 and 1200

    is fixed vertically.

    The velocity of the obstacle is not more than the velocity of the vehicle.

    The vehicle is underactuated and cannot move sideways and roll of the vehicle is

    neglected.

    Length (over all): 4.6m

    Width (over all): 1.6m

    Height: 0.7m

    Propulsion thrusters: 2

    Maneuvering thrusters: 3

    Depth rating:100m

    The positive scaling factors and are taken as 1 and 0.1. The distance influence threshold is

    The forward speed (ud) of the vehicle is fixed to a constant value. This value is used to fix

    the radial distance (dq ) which is used for defining the next one-step positions.

    A sphere of acceptance with radial distance of 2m is considered at goal position.

    Assumptions

    )RR(2d auvobst

  • SP =(0,0,0)

    GP=(90,90,90)

    0

    50

    100

    0

    50

    1000

    20

    40

    60

    80

    100

    Surge(x),[m]Sway(y),[m]

    Hea

    ve(

    z),[

    m] 0 20 40 60 80 100

    0

    50

    100

    Sw

    ay(y

    ),[m

    ]

    Surge(x),[m]

    0 20 40 60 80 1000

    50

    100

    Hea

    ve(

    z),[

    m]

    Sway(y),[m]

    0 20 40 60 80 1000

    50

    100

    Surge(x),[m]

    Hea

    ve(

    z),[

    m]

    3/17/2016 40

    0

    50

    100

    0

    50

    1000

    20

    40

    60

    80

    100

    Surge(x),[m]Sway(y),[m]

    Hea

    ve(

    z),[

    m]

    Vel=2m/s

    Vel=1.5m/s

    Vel=1m/s

    050

    1000

    50

    100

    0

    50

    100

    Surge(x),[m]Sway(y),[m]

    Hea

    ve(

    z),[

    m]

    SP=(10,15,5)

    GP=(83,84,78)

    Local minimum near SP

    0

    50

    100

    0

    50

    100

    0

    50

    100

    Surge(x),[m]Sway(y),[m]

    Hea

    ve(

    z),[

    m]

    0

    50

    100

    0

    50

    100

    0

    50

    100

    Surge(x),[m]Sway(y),[m]

    Hea

    ve(

    z),[

    m]

    Local minimum due to concave shape obstacle

    Simulation results

  • SPPF MPPF

    Distance between obstacle 1 and vehicle (m)

    17.1675 21.9355

    Distance between obstacle 2 and vehicle (m)

    18.1688 22.5124

    Significance of MPPF method

    3/17/2016 41

    0 20 40 60 80 1000

    20

    40

    60

    80

    100

    Surge(x),[m]

    Sw

    ay(y

    ),[m

    ]

    SPPF method

    MPPF method

    0 10 20 30 40 50 60 70 80 90 10040

    50

    60

    70

    80

    Surge(x),[m]

    Sw

    ay(y

    ),[m

    ]

    SPPF

    0 10 20 30 40 50 60 70 80 90 10040

    50

    60

    70

    80

    Surge(x),[m]

    Sw

    ay(y

    ),[m

    ]

    MPPF

    0 20 40 60 80 100 -1000100

    0

    5

    10

    15

    Sway(y),[m]Surge(x),[m]

    SPPF

    Rep

    uls

    ion

    po

    ten

    tial

    0 20 40 60 80 100 -1000100

    0

    5

    10

    15

    Sway(y),[m]Surge(x),[m]

    MPPF

    Rep

    uls

    ion

    po

    ten

    tial

    SPPF MPPF

    Path

    Length

    79 67

    0 5 10 15 20 25 300

    5

    10

    15

    20

    25

    30

    Surge(x),[m]

    Sw

    ay(y

    ),[m

    ]

    D.Fu-guang method

    Proposed method

    Surge(x),[m]

    Sw

    ay(y

    ),[m

    ]

    SPPF

    5 10 15 20 25 30

    5

    10

    15

    20

    25

    30

    Surge(x),[m]

    Sw

    ay(y

    ),[m

    ]

    MPPF

    5 10 15 20 25 30

    5

    10

    15

    20

    25

    30

    0 10 20 30 0102030

    0

    2

    4

    Sway(y),[m]

    SPPF

    Surge(x),[m]

    Rep

    uls

    ion

    po

    ten

    tial

    0 10 20 30 051015202530

    0

    2

    4

    Sway(y),[m]

    MPPF

    Surge(x),[m]

    Rep

    uls

    ion

    po

    ten

    tial

    0

    50

    100

    0

    50

    1000

    50

    100

    Surge(x),[m]Sway(y),[m]

    Hea

    ve(

    z),[

    m]

    0 20 40 60 80 1000

    20

    40

    60

    80

    100

    Surge(x),[m]

    Sw

    ay(y

    ),[m

    ]

    SPPF method

    MPPF method

  • Comparison of MPPF method with other methods

    42

    Surge(x),[m]

    Sw

    ay(y

    ),[m

    ]

    SPPF

    5 10 15 20 25 30

    5

    10

    15

    20

    25

    30

    Surge(x),[m]

    Sw

    ay(y

    ),[m

    ]

    MPPF

    5 10 15 20 25 30

    5

    10

    15

    20

    25

    30

    0 10 20 30 0102030

    0

    2

    4

    Sway(y),[m]

    SPPF

    Surge(x),[m]

    Rep

    uls

    ion

    po

    ten

    tial

    0 10 20 30 051015202530

    0

    2

    4

    Sway(y),[m]

    MPPF

    Surge(x),[m]R

    epu

    lsio

    n p

    ote

    nti

    al

    0 10 20 300

    5

    10

    15

    20

    25

    30

    Surge(x),[m]

    Sw

    ay

    (y),

    [m]

    0 5 10 15 20 25 300

    5

    10

    15

    20

    25

    30

    Surge(x),[m]

    Sw

    ay(y

    ),[m

    ]

    VFF method

    MPPF method

    -15 -10 -5 0 5 10 15-15

    -10

    -5

    0

    5

    10

    15

    Surge(x),[m]

    Sw

    ay

    (y),

    [m]

    SP

    GP

    MPPF

    APF

    NPF

    -15 -10 -5 0 5 10 15-15

    -10

    -5

    0

    5

    10

    15

    GP

    SP

    Surge(x),[m]

    Sw

    ay(y

    ),[m

    ]

    CS boundary

    MPPF method

    NPF method

    APF method

    0 5 10 15 20 25 300

    5

    10

    15

    20

    25

    30

    Surge(x),[m]

    Sw

    ay

    (y),

    [m]

    VFF method

    MPPF method

    Method Local

    minimum

    Distance travelled to

    reach goal position

    APF Yes Couldnt reach

    NPF No 37.9873m

    MPPF No 33.3371m

    Method Chance of

    collision

    Distance travelled to

    reach goal position

    APF Yes 22.36m

    NPF No 37.9873m

    MPPF No 33.3371m

  • 020

    4060

    80100

    120

    -50

    0

    50

    -50

    0

    50

    Surge(x),[m]Sway(y),[m]

    Hea

    ve(

    z),[

    m]

    SP

    GP

    Desired path

    Actual path

    obstacle path

    In MPPF method the distances to the obstacles are increased by 27.8% compared to the conventional method, thus providing a safe path for AUV. Similarly it was observed that the path length is reduced by 17.9%

  • User Input

    Obstacle avoidance and

    way-point guidance

    Control Plane

    Trajectoryplanner

    Sensor data

    Controller

    ThrusterPath

    elements

    Desired

    states

    Tracking

    error

    Speed

    Angle

    Actuators

    Force/TorqueActual states

    +-

    The AUV dynamics can be expressed as:

    where,M=Mass matrix, C()= Coriolis and Centripetal Matrix, D()= Damping matrixg() = Gravitational and buoyancy effects, = Control inputs

    The above equation can be written as, )))g()D()C(((M1

    )(g)(D)(CMBody

    frame

    Inertial

    frame

    x,

    z,

    y,

    p,pK,

    u,uX,

    q,qM,

    v,v,Y

    w,w,Z r,r,N

    AUV Frame assignment

    3/17/2016 44

    Robot Dynamic

    Model

    Integrating high level control with AUV dynamics

  • Simulation results

    3/17/2016 450

    50

    100

    150

    200

    2500

    100

    200

    -100

    0

    100

    Surge(x),[m]Sway(y),[m]

    SP

    WP

    GP

    Desired

    Actual

    0 50 100 1500

    50

    100

    150

    Surge(x),[m]S

    way

    (x),

    [m]

    SP

    GP

    WP

    Desired path

    Actual path

    0 20 40 60 80 100 1200

    100

    200

    Time(t),[s]

    Su

    rge(

    x),

    [m]

    0 20 40 60 80 100 1200

    50

    100

    Time(t),[s]

    Sw

    ay(y

    ),[m

    ]

    0 20 40 60 80 100 120-50

    0

    50

    Time(t),[s]

    Yaw

    (),

    [deg

    ]

    Desired

    Actual

    0 50 100 150

    -50

    0

    50

    -50

    0

    50

    Sway(y),[m]

    Surge,(x),[m]

    Hea

    ve(

    z),[

    m]

    SP

    GP

    Desired Path

    Actual Path

    0

    50

    100

    150

    -100

    -50

    0

    50

    100-50

    0

    50

    Surge(x),[m]Sway(x),[m]

    Hea

    ve(

    z),[

    m]

    SP

    GP

    WP

    Desired path

    Actual path

    Avoiding local minima Combined result

    Obstacle

    avoidance

    3D way-point guidance

    2D way-point guidance

  • Hardware-In-Loop (HIL) Experiments

    dSPACE DS1104 R&D Controller

    GUI

    (ControlDesk)

    Actuator Unit

    (Thrusters and

    Control planes)

    Interface

    Unit

    (Connector

    panel)

    Sens

    or

    User

    Input

    Sensor

    data

    Measured pitch angle

    Measured yaw angle

    Measured

    speed

    Commanded pitch

    angle

    Commanded yaw

    angle

    Commanded

    speed

    Obstacle Avoidan

    ce

    Trajectory

    Planner

    AUV Controll

    er

    AUV Dynami

    cs

    Desired states

    Path element

    s

    Commanded speed & angle

    ForceSP & GP

    dSPACE DS1104 R&D

    controller

    Actualstates Actuato

    r Dynami

    cs

    Measured speed & angle

    3/17/2016 46

    Hardware-In-Loop (HIL) simulation is a technique that is used in the

    development and test of complex real-time systems.

    It replaces the emulated hardware under test or control logic in the

    simulation model, with real hardware that interacts with the computer

    models.

    It provides the realism of the simulation and provides access to

    hardware features currently not available in software-only simulation

    models.HIL architecture

  • 3/17/2016 47

    Thruster

    Rudder motorConnector panel

    Driver boards

    Stern motor

    RTI model

    HIL setup

  • ControlDesk Layout

    3/17/2016SSK,IITM

    48

    HIL test using obstacle sensor

    3DHILcase-1.avi3DHILcase-1.avieditedwosen.avieditedwosen.avi

  • 3/17/2016SSK,IITM

    49

    HIL simulation results

    0 20 40 60 80 100 120-20

    0

    20

    Surge(x),[m]

    Sw

    ay(y

    ),[m

    ]

    -10 -5 0 5 10 15 20-20

    0

    20

    Sway(y),[m]

    Hea

    ve(

    z),[

    m]

    Desired

    Actual

    0 20 40 60 80 100 120-20

    0

    20

    Surge(x),[m]

    Hea

    ve(

    z),[

    m]

    0 50 100 150-5

    0

    5

    Time(t),[s]

    xe,[

    m]

    0 50 100 150-5

    0

    5

    Time(t),[s]

    ye,[

    m]

    0 50 100 150-5

    0

    5

    Time(t),[s]

    z e,[

    m]

    0 50 100 150-20

    0

    20

    Time(t),[s]

    e,[

    deg

    ]

    0 50 100 150-20

    0

    20

    Time(t),[s]

    e,[

    deg

    ]

    Comparison between MIL and HIL

    Error MIL HIL

    xe (m) 0.6018 0.9573

    ye (m) 0.3237 0.3528

    ze (m) 0.3678 0.2187

    Pitche (deg) 2.2558 2.4734

    Yawe (deg) 1.4646 1.8525

    0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

    x 104

    -1

    0

    1x 10

    4

    Xcorr

    ela

    tion

    Lags

    Surge error

    0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

    x 104

    -1000

    0

    1000

    Xcorr

    ela

    tion

    Lags

    Sway error

    0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

    x 104

    -1000

    0

    1000

    Xcorr

    ela

    tion

    Lags

    Heave error

    Error No. of samples

    delayed

    xe (m) 34

    ye (m) 84

    ze (m) 89

    020 40 60 80 100

    -50

    0

    50

    -50

    0

    50

    Sway(y),[m]Surge(x),[m]

    Hea

    ve(

    z),[

    m]

    SP

    GP

    Desired Path

    Actual Path

  • Video

    ../Videos/Case1A_HILdyn3Donline.avi

  • Summary

    Underwater Robots are finding wide applications in Oil and

    gas Industry, Oceanographic explorations and defence

    areas.

    AUVs are the promising robots for the future

    Guidance, Navigation, Control, and power source are the

    present challenges in AUV development

    References1. Fossen, T., 1994. Guidance and Control of Underwater Vehicles. Wiley,

    New York.2. www.rov.org3. D. Richard Blidberg, The Development of Autonomous Underwater

    Vehicles (AUV); A Brief Summary, Autonomous Undersea SystemsInstitute technical report, Lee New Hampshire, USA.

    http://www.rov.org/

  • Thank you