Table 2 Comparison of experiments in two scenes.

From: GMM-searcher: efficient object search in large-scale scenes using large language models

 

Method

First-time search

Repeated search - Same task

Repeated search - Changed task

Path length (m)

Time (s)

Path length (m)

Time (s)

Path length (m)

Time (s)

Avg

Std

Avg

Std

Avg

Std

Avg

Std

Avg

Std

Avg

Std

Scene 1

EFP6

514.43

45.28

573.47

53.77

–

–

–

–

1442.51

141.32

1513.65

122.78

TARE43

665.57

61.45

747.83

58.23

–

–

–

–

1338.67

133.67

1487.41

130.56

SEER54

672.44

57.73

741.15

49.48

–

–

–

–

1424.55

127.82

1489.45

119.88

Ours

426.29

14.67

450.31

12.45

401.78

6.45

409.97

5.83

575.12

21.23

599.08

12.9

Scene 2

EFP6

347.86

33.75

388.53

28.17

–

–

–

–

332.56

31.73

353.54

30.02

TARE43

388.44

32.46

446.48

29.78

–

–

–

–

370.48

33.78

421.23

30.45

SEER54

364.31

31.68

382.16

32.46

–

–

–

–

401.28

38.84

427.51

39.37

Ours

160.63

8.12

167.32

6.34

156.45

3.64

161.29

3.76

201.11

5.12

209.49

4.89

  1. ‘–’ Indicates the metric similar to that in the first-time search, and bold indicates optimal metrics.