Table 1 Generalization performance evaluation in Test Scenario-4

From: Cognitive embodied learning for anomaly active target tracking

Method

No distractor

1 distractor

3 distractors

6 distractors

10 distractors

 

SR(%)

RPL

SR(%)

RPL

SR(%)

RPL

SR(%)

RPL

SR(%)

RPL

NBV

35.6

25.5

18.8

51.8

10.9

102.4

5.3

119.2

-

-

EI (STF)

54.8

15.8

43.7

30.1

35.6

49.8

26.4

101.4

17.1

125.8

EI (SI)

69.5

15.7

62.0

19.7

56.0

22.6

50.0

26.3

41.7

30.1

EI (EA)

77.9

14.3

70.6

15.4

64.8

16.0

57.4

16.7

49.6

17.4

EI (EA+SI)

81.4

12.1

75.8

12.8

71.7

13.2

66.5

13.7

60.4

14.3

CEL (Rules)

91.1

8.1

89.5

8.6

85.6

8.9

81.2

9.5

76.9

10.1

  1. The values in bold indicate the best-performing results in every setting. SR is the success rate, and RPL is the relative path length. EL(EA + SI), EL(EA), EL(SI), and EL(STF) represent four common embodied learning (EL) paradigms. EA+SI is a policy learning method that uses the semantic feature cognition (SFC) and the sequential cognitive memory (SCM) to process observations. EA is a policy learning method that uses the SFC to process observations. SI is a transformer-based method that uses the SFC and the SCM to generate semantic tokens. STF is an adaptive embodied learning method based on the student-teacher framework. NBV is a region-based viewpoint optimization computational method.
Back to article page