2 3 4 Extension of the presentation at ARMS@AAMAS 2015 5 6 7 8 - - PowerPoint PPT Presentation

2 3 4 extension of the presentation at arms aamas 2015 5
SMART_READER_LITE
LIVE PREVIEW

2 3 4 Extension of the presentation at ARMS@AAMAS 2015 5 6 7 8 - - PowerPoint PPT Presentation

Source: robocup.org 2 3 4 Extension of the presentation at ARMS@AAMAS 2015 5 6 7 8 Compute new deployment t 1 t 1 on graph G ( V , C ) iff agent a in vertex v, z 1 av in this case collect


slide-1
SLIDE 1
slide-2
SLIDE 2

2

Source: robocup.org

slide-3
SLIDE 3

3

slide-4
SLIDE 4

4 Extension of the presentation at ARMS@AAMAS 2015

slide-5
SLIDE 5

5

slide-6
SLIDE 6

6

slide-7
SLIDE 7

7

slide-8
SLIDE 8

8

slide-9
SLIDE 9

9

  • Compute new deployment
  • n graph
  • iff agent a in vertex v,

in this case collect utility proportional to information gain and distance to v

  • Constraints (5) exponential

in number, but model can be solved optimally

1 

av

z ) , (

1 1  

t t

C V G

slide-10
SLIDE 10

10

Method Communication constraints Coordination Optimal Hard Hard Centralized Stump [Stump et al., 2011, ICRA] (adapted) Hard Centralized Rooker [Rooker and Birk, 2007, Control Eng Pract] Hard (continuous) Centralized Utility [Spirin et al., 2013, TAROS] r  0 greedy, r  1 quasi-hard r=0.1,0.5,0.9 Soft Decentralized

slide-11
SLIDE 11

11

slide-12
SLIDE 12

12

Office Maze Open

slide-13
SLIDE 13

13

Office – instant replan

slide-14
SLIDE 14

14

Office – with repl. time (1 step = 2 seconds)

slide-15
SLIDE 15

15

Maze – instant replan

slide-16
SLIDE 16

17

Optimal Hard Utility 0.5

slide-17
SLIDE 17

18

Open – instant replan

slide-18
SLIDE 18

20

slide-19
SLIDE 19

21