版权说明:本文档由用户提供并上传,收益归属内容提供方,若内容存在侵权,请进行举报或认领
文档简介
Aeronautics
&
AstronautAutonomous
Flight
SystemsLaboratoryAll
slides
and
material
copyright
ofUniversity
of
Washington
AutonomousFlight
Systems
LaboratoryAeronautics
&
AstronautAutonomous
Flight
SystemsLaboratoryResearch
and
Development
at
theAutonomous
Flight
Systems
LaboratoryUniversity
of
WashingtonSeattle,
WAGuggenheim
109,
AERB
214(206)
543-7748/research/afslReal
Time
Strategic
Mission
PlanningBaseObstacle/ThreatAvoidanceSearching/Target
IDCoordination
w/
surface
vehiclesAutonomous
Flight
Systems
Laboratory
Aeronautics
&
AstronautiPattern
hold/Team
assembly
TransitionUniversity
of
Washington3Aeronautics
&
AstronautiAutonomous
Flight
Systems
LaboratorySystem
OverviewPreviously
funded
by
DARPA
&
AFOSRUniversity
of
Washington4Aeronautics
&
AstronautiAutonomous
Flight
Systems
LaboratorySystem
Block
DiagramSolving
optimal
control
problems
in
real-timeUniversity
of
Washington5Stochastic
Problem
FormulationAutonomous
Flight
Systems
Laboratory
Aeronautics
&
AstronautiPredicted
probability
of
survival
of
each
vehicle
at
time
tq+1Predicted
probability
that
a
task
is
not
completedat
time
tq+1Team
utility
functionUniversity
of
Washington6Add
stuff
about
BAT
SIMDistributed
Architecture
forCoordination
of
Autonomous
VehiclesAutonomous
Flight
Systems
Laboratory
Aeronautics
&
Astronauti
Each
vehicle
plans
its
ownpath
and
makes
task
tradingdecisions
to
maximize
theteam
utility
function
There
is
one
activecoordinator
agent
at
a
timeefficiencyfailure
detection
local/global
informationexchanges
Computational
requirementfor
running
coordinator
agentis
small
compared
to
planning
Coordinator
role
can
betransferred
to
another
vehiclevia
a
voting
procedureUniversity
of
Washington7Evolution-based
Cooperative
PlanningSystem
(ECoPS)Autonomous
Flight
Systems
Laboratory
Aeronautics
&
Astronauti
Uses
Evolutionary
Computation-based
techniques
in
theoptimization
of
trading
decisionmakingandpath
planning
Task
planner
uses
price
andshared
information
in
addition
topredicted
states
of
the
world
formaking
trading
decisions
Task
planner
interacts
with
pathplanner
and
state
predictor
tosimultaneously
search
feasiblenear-optimal
task
and
path
plans.
We
call
this
system
the“Evolution-Based
CollaborativePlanning
System”
–
ECoPS,combining
market
basedtechniques
with
evolutionarycomputation(EC).University
of
Washington8Aeronautics
&
AstronautiAutonomous
Flight
Systems
LaboratoryEvolutionary
Computation
(EC)
Motivated
by
evolutionprocess
found
in
nature
Population-basedstochastic
optimizationtechniqueMetaphorMappingUniversity
of
Washington9Aeronautics
&
AstronautiUniversity
of
Washington10Features
of
Evolution-BasedComputationAutonomous
Flight
Systems
LaboratoryProvides
a
feasible
solution
at
any
timeOptimality
is
a
bonusDynamic
replanningNon-linear
performance
functionCollision
avoidanceConstraints
on
vehicle
capabilitiesHandling
loss
of
vehiclesOperating
in
uncertain
dynamic
environmentsTiming
constraintsAeronautics
&
AstronautiMarket-based
Planning
forCoordinating
Team
TasksAutonomous
Flight
Systems
LaboratoryTask
allocation
problem:At
trading
round
nThe
goal
of
task
trading:Each
vehicle
proposeswhich
are
approved
by
the
auctioneerbased
on
bid
price.At
the
end
of
the
trading
round:University
of
Washington11Distributed
Task
Planning
AlgorithmAeronautics
&
AstronautiAutonomous
Flight
Systems
LaboratoryDynamic
Path
Planning
Generate
feasible
paths
andplanned
actions
within
aspecified
time
limit
(ΔTs
)while
the
vehicles
are
inmotion.
Highly
dynamic
environmentrequires
a
high
bandwidthplanning
system
(i.e.small
ΔTs).
Formulate
the
problem
as
aModel-basedPredictiveControl
(MPC)
problemUniversity
of
Washington12Aeronautics
&
AstronautiAutonomous
Flight
Systems
LaboratoryEC-Based
Path
PlanningMutationDynamic
PlanningPath
EncodingUniversity
of
Washington13Aeronautics
&
AstronautiAutonomous
Flight
Systems
LaboratoryCollision
Avoidance
Model
each
site
in
the
environment
as
auncertainty
circular
area
with
radiusProbability
of
intersection:use
numerical
approximationcomputationally
easier
than
true
solution:
possible
intersection
region:
probability
density
field
function:
position
on
the
path
Ci
:
expected
site
locationv
:
velocity
of
the
vehicleUniversity
of
Washington14Aeronautics
&
AstronautiAutonomous
Flight
Systems
LaboratoryCollision
Avoidance
ExampleUniversity
of
Washington15Aeronautics
&
AstronautiAutonomous
Flight
Systems
LaboratorySimulation
ResultsSimulation
on
the
Boeing
Open
Experimental
PlatformUniversity
of
Washington16Some
Aspects
of
ECoPSUniversity
of
Washington17Autonomous
Flight
Systems
Laboratory
Aeronautics
&
Astronauti
Eachvehicle
computes
its
own
trajectory
and
makes
decisionto
trade
its
tasks
with
other
vehicles.Vehicles
may
sacrifice
themselves
if
that
benefits
the
team.
Each
vehicle
needs
to
have
periodically
updated
locations
ofnearby
vehicles
only
for
collision
avoidance.
Each
vehicle
needs
to
know
the
information
about
theenvironment.
The
accuracy
of
the
information
affects
thequality
of
its
decision
making.
The
rate
of
environment
information
updates
shou
温馨提示
- 1. 本站所有资源如无特殊说明,都需要本地电脑安装OFFICE2007和PDF阅读器。图纸软件为CAD,CAXA,PROE,UG,SolidWorks等.压缩文件请下载最新的WinRAR软件解压。
- 2. 本站的文档不包含任何第三方提供的附件图纸等,如果需要附件,请联系上传者。文件的所有权益归上传用户所有。
- 3. 本站RAR压缩包中若带图纸,网页内容里面会有图纸预览,若没有图纸预览就没有图纸。
- 4. 未经权益所有人同意不得将文件中的内容挪作商业或盈利用途。
- 5. 人人文库网仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对用户上传分享的文档内容本身不做任何修改或编辑,并不能对任何下载内容负责。
- 6. 下载文件中如有侵权或不适当内容,请与我们联系,我们立即纠正。
- 7. 本站不保证下载资源的准确性、安全性和完整性, 同时也不承担用户因使用这些下载资源对自己和他人造成任何形式的伤害或损失。
最新文档
- 二氧化碳分析仪项目可行性分析报告范文
- 咨询顾问公司面试技巧及题目详解
- 宝武集团项目经理面试题库含答案
- 深度解析(2026)《GBT 18946-2003橡胶涂覆织物 橡胶与织物粘合强度的测定 直接拉力法》(2026年)深度解析
- 品牌经理岗位面试题及市场分析能力含答案
- 供水设备建设项目可行性分析报告(总投资5000万元)
- 石油化工设备工程师面试要点与答案
- 建筑设计师岗位的面试题及答案
- 物资出入库自动化管理方案
- 珠宝销售面试题及答案
- 结构加固施工验收方案
- 小班美术活动《漂亮的帽子》课件
- 矿山破碎设备安全操作规程
- 暖通工程调试及试运行总结报告
- 2024年全国职业院校技能大赛ZZ054 智慧物流作业赛项赛题第2套
- 《药品质量管理体系内审员职业技能规范》
- 冶炼厂拆迁施工方案
- 谷物烘干机结构设计
- 新疆交通投资责任有限公司 笔试内容
- 检修安全培训内容课件
- 颅内感染指南解读
评论
0/150
提交评论