• No results found

Reasoning About Commitments and Penalties for Coordination Between Autonomous Agents

N/A
N/A
Protected

Academic year: 2020

Share "Reasoning About Commitments and Penalties for Coordination Between Autonomous Agents"

Copied!
8
0
0

Loading.... (view fulltext now)

Full text

(1)

Reasoning about Commitments and Penalties for

Coordination between Autonomous Agents

Cora B. Excelente-Toledo

University of Southampton

Dept. of Electronics and

Computer Science

Southampton SO17 1BJ, UK

cbet99r@ecs.soton.ac.uk

Rachel A. Bourne

Queen Mary

Univ. of London

Dept. of Elec. Eng.

London E1 4NS, UK

r.a.bourne@elec.qmw.ac.uk

Nicholas R. Jennings

University of Southampton

Dept. of Electronics and

Computer Science

Southampton SO17 1BJ, UK

nrj@ecs.soton.ac.uk

ABSTRACT

1. INTRODUCTION

!

"

#

$

%

#

&

'()*

Permission to make digital or hard copies of all or part of this work for

personal or classroom use is granted without fee provided that copies are

not made or distributed for pro

fi

t or commercial advantage and that copies

bear this notice and the full citation on the

fi

rst page. To copy otherwise, to

republish, to post on servers or to redistribute to lists, requires prior speci

fi

c

permission and/or a fee.

AGENTS’01,

May 28-June 1, 2001, Montr´eal, Quebec, Canada.

Copyright 2001 ACM 1-58113-326-X/01/0005 ...

+

5.00.

,

-

.

"

'/0)

1 2

1'3)

451'/6) !

%

'7) "

#

'(),

%

Æ &

'0(8) "

9

Æ

#

'//) 9

(2)

: #

;

-

5

<

#

&

# 5

;

'7)

;

<

;7

;=

;0

;6

;3;(

2. THE SCENARIO

!

> .

; ,

4 .

24

;

!

;

4 /

!

; 4

9 ;

/ 7 = 0 6

/

7

=

0

6

9 4

; 4

4< 9

4< .4<

*

& 4<

4

?

!

4<

$4

!4!4

4<

4

4

9 -

!4!4

!4 4 !

; !;

;

7

9 '7)

4

4 4

9 4

!

% 4

!!4 !4

4

9

!4 9

!4!4

&

!4 4

(3)

! " #$

% &

' (

(#$&

)

)

# (

( $

*

(&

( ( (

+ , -

(

! 4

!4

4 !4

9

!4

@!4

!4

!4!4

4 4

4

3. COMMITMENTS AND PENALTIES

!4 !4 9

!49

!4

4

#

9

*

*

4

A* !

1* ! 4

4 &

6BC

4

!

4 1

!406

,*

&*

1*

!4-

;*

4 ,

$

9 4

!4

4

4. AGENT DECISION MAKING

'7)

?

,?

,

9

;

,

'D)

*

, 4< ,

,

,

4.1 Deciding on the direction of movement

(4)

E E

4.2 Deciding which CM to select

!

4

4<

4

,

4

4

4<

4<

,

!4!4

94

!4!;

4

4<

4<

? ; 4

F

9

?

4

G

4 ' )

F

&

4<

;!4

*

F

H

; !4

@

*

F

4

4

F

I !4

*

F

H

54< ?

"4<

4< 9

4<

4<

G /

!!;

F/B 66

4 F3

'=7) '=7)73

;

=7 7 2

FB376

F=B8 *

=7 F B83/B =6H/=D8

F B6//

4.3 Deciding what to bid to become an AiCoop

" 4

4

!

4 5

4

- ? 4

?

!

/ '6=) 0

'70) 3

'=7) 7 % '//)

( '06)

( '=7),

B

9 %

(5)

4

4<

*

F

9!4!;

4.4 Deciding which AiCoop bids to accept

!4

&

!4

;!4

4 !4

!4 4 !4

&

&

!4

&!4

!4

*

F

H H

5 *

F' )

&

¾Ë ')

*

F

¾Ë

' )HH

¾Ë ')

5

4<

Æ 9

!4;

4.5 Deciding how to set the penalty fee

9

4

F

9 &

4 9

,

!4

F/BBC

4 9

4

4

F

1

4.6 Deciding when to drop a commitment

* !4

!4 !4

!4

94

<

!!4

!44 *

;!4

!44 *

5. EXPERIMENTAL EVALUATION

*

*

/BBBB ? /B/B 6

= 4

? 0

F// ; /4

'7B 0B 3B)

B76

6B (6/BBC

4

4

/B 9

! !5J!

$

BB6 'BBBB=()

(6)

0

500

1000

1500

2000

2500

3000

Total

Partial (50%)

Loose

Level of Commitment

Tot

al reward obt

ained by all agent

s

0

200

400

600

800

1000

1200

1400

1600

Number of CTs achi

eved

AiC

AiCoop

AiS

CTs accomplished

&

76C4 &=

;

%

,

4

#

=4

%

4 ,!4

9!4!;

!

-!4

4 07

0=

44

!

& 0

4 !4

, !4

76C

(6C 9 !4

BC

0

10

20

30

40

50

60

70

0%

25%

50%

75%

100%

Percentage of Commitment

Nu

m

ber

of

C

Ts decom

m

itt

ed

Unrecovered

Recovered

0

10

20

30

40

50

60

0%

25%

50%

75%

100%

Fixed Penalty with a percentage of the CT Reward

Number

of CTs decommitted

0

5000

10000

15000

20000

25000

Aver

age penalty fee

Recovered

Unrecovered

Penalty Fee

!

0

5

10

15

20

25

30

35

40

0%

25%

50%

75%

100%

Level of Commitment

Nu

m

ber

of

C

Ts decom

m

itt

ed

0

100

200

300

400

500

600

700

800

900

Aver

age penal

ty f

ee

Recovered

Unrecovered

Penalty Fee

(7)

0

5

10

15

20

25

30

35

40

0%

25%

50%

75%

100%

Level of Commitment

Nu

mb

er

o

f CT

s d

ec

ommi

tte

d

0

100

200

300

400

500

600

700

800

900

Average penalty fee

Recovered

Unrecovered

Penalty Fee

# $%

0

500

1000

1500

2000

2500

3000

3500

Total Partial

(50%)

Loose

Level of Commitment

AiC Total reward

Sunk Cost

Partially Sanctioned

Penalty Fixed 50% CT

Penalty Fixed 0% CT

& '!(

%

&63

(

"6

!4 !4

6BC

"3

(

%

K

9

!4

D

!4

6BC "

BC

6BC !4

6BC

I

9

#

6. RELATED WORK

*

, .

9

'/0/63/B)

:'6)

#

;#

L'() %

%

G

'/)

%

G'=)

<

%

9

; A-

'//)

/ ;-

$ %

&

$ ;

(8)

'/7) 9

5

;:-

'/=)

MN

7. CONCLUSIONS

# 9

#

"

$ &

&

-

#

8. ACKNOWLEDGMENTS

<-5

4; 45!4

9. REFERENCES

'/) O;G:4% %A4

9

/P/6<

!7BBB

L>

9

!"#

$"#%&&&'=0DP=67!7BBB

'=) 4G;

9(

)#

$)#**'0(DP0D6;

;/888

'0) 14%LA 9

07*7/=P73//88B

'6) .%:1+

7B/ *88P//3 /888

'3) .%:J>A1

* !

"""( ,

,#7/6 *//3(P//D= /88/

'() 5>L4 *

-" D= *77=P76B

/88=

'D) : 5O<124

9

)#

$)#* 'D7PDD/88/

'8) %> %

I 1/88(

'/B) L;>2Q"

<9 1* 4<!/880

'//) ";J>A!

9

. #

$*/'/73P/==

1!/883

'/7) ";;;;5!

? 9

()

# $)#**'

6=6P60B;;/888

'/=) ;;.%:

)

0# (

=/ *3(PD/ /880

'/0) K;< ?

9

.#

$*%'7(3P7D/;

L4 L/887

'/6) >2;>: &

"""

( ,,#

References

Related documents

In the next sections I propose a few preliminary steps- first, in section II, I explain how, at least in Kantian theory, particular facts become reasons to act insofar as they

The main purpose of this paper is to analyze the causal relationships between defence spending and economic growth using the Toda–Yamamoto approach to Granger causality test in

Molecular Weight Distribution, Carboxylic Acidity and Humic Substances Content of Aquatic Organic Matter: Implications for Removal During Water Treatment. Removal of Organic Matter

However, the study found that the child protection strategies had achievement during implementation of child protection project in Magu.30.6 percent of respondents

The higher expression level of SbK + O , coding for an osmotic stress potassium transmembrane transporter, recorded in the drought tolerant IS22330 compared to the sensitive

Odds of Complying With Blood Stool Test and Prior Use of Colonoscopy Among Households Containing at Least 1 Member Age 50 Years or Older in Little Haiti, Miami-Dade County,

We will follow the path of the devel- opment, from the design, analysis, and implementation of McEliece cryptosystem and the quantum attack resistance to the latest fully

If qubit 3 is affected by the time-correlated error, bit-flip, phase-flip or both, the disentanglement circuit will propagate these errors to the extra ancilla qubits: a CNOT gate