• No results found

Goal driven planning and adaptivity for AUVs CAR06

N/A
N/A
Protected

Academic year: 2021

Share "Goal driven planning and adaptivity for AUVs CAR06"

Copied!
44
0
0

Loading.... (view fulltext now)

Full text

(1)

Goal driven planning and adaptivity for AUVs

(2)

GESMA : Location

(3)

Underwater Warfare

Environment

Platform susceptibility

Robotics

Mine warfare

GESMA : Skills & Missions

(4)

GESMA

(5)

DCSD Lab:

Systems Control and Flight Dynamics Department

execution control

situation

assessment

planning

guidance

state

estimation

operator

operator

Flight mechanics

& Identification

Control

Decision

Human Factors

vehicle

vehicle

ONERA :

(6)

PROLEXIA

PROLEXIA

Small company

Simulation

(7)

Gesma studies on planning tools

Gesma studies on planning tools

Assess requirements for mission planning

tools for different levels of autonomy

Assess adaptivity for Mine Warfare and

(8)

Levels of autonomy

Levels of autonomy

Level 1: Ordered set of elementary controls

Level 2 : Geographic trajectory planning (Wpts)

Level 3 : Goal driven planning & Operator supervision

Level 4 : Fully autonomous mission

(9)

Military

Military

requirements

requirements

•Low bandwidth communication

•Discrete or Unsupervised AUV mission

•Unknown or hostile environment

Adaptivity

Autonomy

(10)

Examples

Examples

of

of

goals

goals

Find a wreck (or plane black boxes)

Survey a zone for mines detection,

classification and identification

Find a free path suitable for amphibious

(11)

Examples

Examples

of

of

adaptivity

adaptivity

If the current is strong and not in the main survey direction,

change the survey direction

If something is detected, try to classify by multi-aspect sonar

acquisition

If something is classify, try to identify by going over at low

altitude

If enter a posidony field, change mapping strategy

If sand ripples prevent from good detection, plan another

(12)

Mission planning

Mission planning

Data quality insurance

Exhaustivity (coverage, overlap, redundancy)

Accuracy (altitude related to sensor definition, environment)

Confidence (computer aided decision, performance criteria,

navigation errors)

AUV security insurance

Bathymetry

Forbidden area

Security immersion in traffic zone

Strong currents

Mission optimization

Mainly for energy consumption

Time/tide/currents

(13)

What you need is:

What you need is:

A nice AUV

Redermor platform

User friendly Operator MMI

Preparation / Supervision

Environmental database

SHOM METOC guide / C-Map

Automatic planning algorithm

Dijkstra and Little algorithms (shortest path, lowest cost)

Onboard supervision system

Petri nets

Events generation by sensors measurements and computer

(14)

The nice AUV :

The nice AUV :

Redermor

Redermor

Sensors Planning optimization PLN Petri player ProCoSa Data manager GDD Petri nets Vehicle behavior Guidance GUI EVT Interface of commands IDC

Data server of GESMA OA_NAVIO

Payload drivers and

treatments

CAN Bus

Drivers Drivers Actuators Can big boxes OA1

OA2 mission

OA2 and OA3

consigns events commands status events Sensors Planning optimization PLN Petri player ProCoSa Data manager GDD Petri nets Vehicle behavior Guidance GUI EVT Interface of commands IDC

Data server of GESMA OA_NAVIO

Payload drivers and

treatments

CAN Bus

Drivers Drivers Actuators Can big boxes OA1

OA2 mission

OA2 and OA3

consigns

events commands

status events

(15)

Mission planning MMI

Mission planning MMI

level 1 : text editor Or NIVAS/IOVAS

level 2 : FLEET MANAGER

(16)

Environmental database

(17)

Planning algorithms

Planning algorithms

mission graph waypoint mission graph waypoint obstacles -100 -80 -60 -40 -20 0 0 10000 20000 30000 40000 50000 60000 survey 1 survey 2 survey 3 obstacle avoidance

(18)

Supervision by Petri nets

(19)

Events generation, data treatment

Events generation, data treatment

Apply to mine warfare

événements + informations

actions = requête de calcul

vers rdp ou serveur

événements + résultats

Re planning

(20)

Level 1 : IOVAS MMI (

(21)

Level 1 : sequence of controls

(22)

Level 1 : mission script

(23)

Paramètres Type de commande du ré e v it e s s e _ a v a n c e im m e rs io n _ fi n a le v it e s s e _ c h g t_ im m e rs io n a lt it u d e fo n d _ fi n a le v it e s s e _ c h g t_ a lt it u d e fo n d d e lt a _ im m e rs io n d e lt a _ a lt it u d e fo n d v it e s s e _ g ir a ti o n c a p _ fi n a l d e lt a _ c a p SuiviCapVitImm x SuiviCapVitAltifond x SuiviCapImm x x SuiviCapAltifond x x SuiviCapVitImmersionFinale x x SuiviCapVitAltitudefondFinale x x SuiviCapVitDeltaImmersion x x SuiviCapVitDeltaAltitudefond x x GirationImm x x GirationAltifond x x GirationImmCapFinal x x GirationAltifondCapFinal x x GirationImmDeltaCap x x GirationAltifondDeltaCap x x GirationImmersionFinale x x x GirationAltitudefondFinale x x x GirationDeltaImmersion x x x GirationDeltaAltitudefond x x x

Level 1 : controls

Level 1 : controls

(24)

Level 1: supervision

(25)

Level 2 MMI : Fleet manager (ACSA)

(26)

Level 2: what an AUV can do?

(27)

Level 2 : waypoints definition

(28)

Level 2 : mission script

(29)

Level 2: mission supervision

(30)

Level 2 : mission example

(31)

Level 2 : mission script example

(32)

Level 3 : IOVAS MMI (

(33)

Level 3 : IOVAS MMI

(34)

Level 3 : interaction with environment

(35)

Level 3 : interaction with environment

(36)

Level 3 : Mission preparation

(37)

Level 3: Missions goals

(38)

Level 3: Automatic planning

(39)

Level 3: Automatic planning

(40)

Level 3: mission script?

(41)

On

On

-

-

going test

going test

Level 1 and 2 :

fully operational

Level 3:

tested in simulation,

at sea this year.

Adaptivity to be tested :

Wrong current direction in the initial planning of a survey

zone

(42)

On

On

-

-

going test

going test

(43)

Further works

Further works

Mine warfare: do sonar acquisition of the

survey zone until the performance criteria

is over a define threshold

REA : Survey strategy with regards to sea

bottom characteristics

Optimization of long duration mission

(44)

Conclusion

Conclusion

Petri nets are a good solution for

supervision and adaptivity

Difficulty of environmental database

integration

Planning algorithm seemed not so

References

Related documents