Skip to main content

Table 1 Average Sensor Alignment over Segments

From: Alignment of angular velocity sensors for a vestibular prosthesis

 

Slow ROM

Fast ROM

Slow Explore

Fast Explore

All

HEAD: HMMS

RMS error

5.4°/s

12.0°/s

15.6°/s

34.2°/s

17.0°/s

Point-to-point error

61.8%

36.8%

108.6%

158.5%

106.6%

Correlation (R2)

0.9364

0.8748

0.7143

0.5100

0.6604

RMS error

33.27%

53.05%

68.86%

78.55%

66.22%

Point-to-point error

42.28%

60.91%

71.22%

86.74%

66.07%

Correlation (R 2 )

124.15%

124.74%

124.16%

127.57%

123.49%

BBS: HMMS

RMS error

4.7°/s

6.4°/s

7.5°/s

17.1°/s

8.7°/s

Point-to-point error

34.9%

40.0%

48.0%

94.1%

54.4%

Correlation (R2)

0.9676

0.9809

0.9391

0.8551

0.9259

RMS error

26.70%

29.50%

38.42%

56.87%

40.09%

Point-to-point error

24.98%

30.69%

39.27%

60.40%

40.01%

Correlation (R 2 )

130.21%

127.48%

126.64%

129.71%

129.26%

Movement description

RMS angular velocity

22.8

34.2

31.0

47.7

33.1

RMS range

45.8

68.4

62.1

95.4

66.2

Range of velocities

233.0

434.0

191.6

292.2

467.2

  1. Alignments are calculated within each segment (or over all segments for All) for each trial by randomly selecting 70% of the data in each trial for training (20 random selections) and using the remaining 30% of data for testing. Metrics were averaged over these 20 test sets, then over the three axes, then over the five trials. Italics entries are the aligned metrics divided by the unaligned measurements for the same segments. For example, HEAD: HMMS RMS error is reduced more than 75% by using the alignment algorithm compared to leaving the sensors unaligned.