Control of Torque-Assisted Bicycle based on Physical Activity during - - PowerPoint PPT Presentation

Control of Torque-Assisted Bicycle based on Physical Activity during Repetitive Prolonged Cycling Exercise

Tohru KIRYU*, Shinya KATO*, Takao MORIYA**, and Yasufumi MIZUNO**

*Graduate School of Science and Technology, Niigata University, 8050 Ikarashi-2, Niigata 950-2181, Japan, **Yamaha Motor Co., Ltd., 2500 Shingai, Iwata, Shizuoka 438-0026, Japan

Control of Torque-Assisted Bicycle based on Physical Activity during Repetitive Prolonged Cycling Exercise, ISEK2004 in Boston, June 19, 2004

• T. Kiryu, Niigata University

Introduction

Power-Assisted Bicycle is now spreading

Easy to climb at steep slope like cycling at flat road

Numbers of Power-Assisted Bicycles in Japan

5 10 15 20 25 93 94 95 96 97 98 99 00 01 02 03

year [ten thousand]

Appropriate power assist based on physical activity is required

Merit: Facilitate voluntary movement and support degenerate muscle force due to aging or small amount of exercise Demerit: excessive assist reduces amount of exercise, while too small assist causes overload for degenerate muscle force

Control of Torque-Assisted Bicycle based on Physical Activity during Repetitive Prolonged Cycling Exercise, ISEK2004 in Boston, June 19, 2004

• T. Kiryu, Niigata University

Background and Purpose

Current power-assisted bicycle is controlled by a given torque from a rider

• Weak point of torque control: no reflection of

physical activity including muscle fatigue of a rider.

• Ex. Reduction of pedal torque due to muscle

fatigue decreases assist torque at the same time. This will induce progression of fatigue.

Assist System

pedal torque driving torque

• Reflection of physical activity on power-assist control is required during cycling

step 1: estimation of physical activity from measured biosignals. step 2: determination of appropriate power-assist .

Physical Activity Determination of Assist Torque Biosignals

step 1 step 2

Control of Torque-Assisted Bicycle based on Physical Activity during Repetitive Prolonged Cycling Exercise, ISEK2004 in Boston, June 19, 2004

• T. Kiryu, Niigata University

Outline

• 1. Measurement of biosignals during cycling with power-assisted bicycle and

evaluation of the correlation between physical activity and power-assist with the progression of fatigue.

• 2. Designing how to control appropriate power-assist based on the physical

activity during cycling.

• 3. Feasibility experiments for a power-assist bicycle equipped with a new

control system.

RSA MPF modeling control by PLC

10 20 30 40 50 start rest start climbing finish finish rest RSA-ratio[%]

• ld syste

new system

Control of Torque-Assisted Bicycle based on Physical Activity during Repetitive Prolonged Cycling Exercise, ISEK2004 in Boston, June 19, 2004

• T. Kiryu, Niigata University

Measurement of Vehicle Data

YAMAHA PAS PX20

• torque assist
• 20 inch wheel
• auto-transmission
• speed limit

driving torque (pedal and assist)

2 Vehicle Data

• speed [m/sec] at the wheel
• cadence [rpm] at the pedal
• pedal torque [Nm]

60 [rpm] was requested by a tone pace maker

magnetic sensors Assist System

pedal torque

speed assist ratio

1 Limit-1 Limit-2

http://www.jbpi.or.jp/english/b.news/enews2.html.

Control of Torque-Assisted Bicycle based on Physical Activity during Repetitive Prolonged Cycling Exercise, ISEK2004 in Boston, June 19, 2004

• T. Kiryu, Niigata University
• ECG gain: 46 [dB]

time constant: 150 [msec]

• EMG gain: 60 [dB]

time constant: 30 [msec] Data Acquisition 12 [bit], sampling frequency

• f 5000 [Hz]

Measurement Conditions

Measurement of Biosignals

( ) ( ) ( )

• =

H L H L

f f t f f t

df f F df f F f t MPF

• +

=

T T

dt t S T t ARV ) ( 2 1 ) (

• fH: low frequencyFt: fft_result

fL: high frequency f: frequency

Signal processing

• ECG
• EMG at right/left vastus lateralis
• Subjective index

Biosignals

R-R interval time-series [sec] MPF: mean power frequency [Hz], ARV: averaged rectified value [µV] NASA-TLX

Control of Torque-Assisted Bicycle based on Physical Activity during Repetitive Prolonged Cycling Exercise, ISEK2004 in Boston, June 19, 2004

• T. Kiryu, Niigata University

Five or six trials for each experimental set measurement

start finish

resting

NASA-TLX 2 [min] 3 [min]

830 [m] long

[min] 5 10 15 20 25 30

1st trial 2nd trial 3rd trial 4th trial 5th trial 6th trial

Route and Time Table

start period climbing period finish period

top view

distance height start period finish period climbing period

side view

237 507

• 6.3˚

3.4˚ 5.7˚

• 2.5˚

[m]

Control of Torque-Assisted Bicycle based on Physical Activity during Repetitive Prolonged Cycling Exercise, ISEK2004 in Boston, June 19, 2004

• T. Kiryu, Niigata University

Two Types appearing in HRV

0.2 0.4 0.6 0.8 1.0 200 250 300 50 20 40 60 50 100 150 200 250 300 start finish 2nd corner 3rd corner time [sec] [sec] [%] time [sec] 100 150 0.2 0.4 0.6 0.8 1.0 50 150 200 250 300 20 40 60 50 100 150 200 250 300 start finish 2nd corner 3rd corner time [sec] [sec] [%] time [sec] 100

GDRR SDRR RSA_ratio RR_interval RSA_ratio RR_interval

distance height start period finish period climbing period side view 237 507

• 6.3˚

3.4˚ 5.7˚

• 2.5˚

[m]

Representation of the overall behavior as a function of distance from the starting point to compare the results for each phase among trials

Control of Torque-Assisted Bicycle based on Physical Activity during Repetitive Prolonged Cycling Exercise, ISEK2004 in Boston, June 19, 2004

• T. Kiryu, Niigata University

RSA: Respiratory Sinus Arrhythmia

• Frequency band of 0.15 [Hz] ~ 0.5 [Hz] in heart rate variability
• Improving gas exchange
• Reflection of sympathetic and parasympathetic nervous activities
• Increase during rest, while decrease under fatigue or stress.
• =

WH WL

f f

dfdt t f W dfdt t f W t ratio RSA ) , ( ) , ( ) ( _

5 . 3 .

f: frequency W: wavelet result fWH: max frequency fWL: min frequency

RSA-ratio

Autonomic Regulation

0.5 [Hz] time

t

0.3

start point climbing period

frequency

Control of Torque-Assisted Bicycle based on Physical Activity during Repetitive Prolonged Cycling Exercise, ISEK2004 in Boston, June 19, 2004

• T. Kiryu, Niigata University
• SDRR (Steeply Decreasing in RR interval before climbing)

mean of RSA-ratio at start period is greater than 20%

• GDRR (Gradually Decreasing in RR interval)
• therwise

Change in physical activity during an experimental set even in the same subject

Grouping by RSA ratio

GDRR and SDRR show the condition of current physical activity and does not show the physical work capacity of each subject Grouping two types of physical activities at each trial by RSA-ratio

Control of Torque-Assisted Bicycle based on Physical Activity during Repetitive Prolonged Cycling Exercise, ISEK2004 in Boston, June 19, 2004

• T. Kiryu, Niigata University

Grouping Results by RSA ratio

• SDRR was dominant for assist-on trials, whereas GDRR for assist-off trials.

power assist might be effective

• However,

there were GDRR even in assist-on trials. power assist is not a definite factor for increasing RSA-ratio

assist SDRR GDRR total ON 54(66.7%) 27(33.3%) 81 OFF 7(31.8%) 15(68.2%) 22 total 61(59.2%) 42(40.8%) 103

Subjects: healthy 8 male and 5 female students (21–24 yrs. Old) total 103 trials 22 trials were assist-off trials) Conventional power assist strategy is not suitable for rider’s physical work capacity or conditions

Control of Torque-Assisted Bicycle based on Physical Activity during Repetitive Prolonged Cycling Exercise, ISEK2004 in Boston, June 19, 2004

• T. Kiryu, Niigata University

5 10 15 20 25 30 35 5 10 15 20 25 30 35 phases 1 and 2 phases 3 and 4 phases 5 and 6

[%]

preceding rest following rest p < 0.05

* * * * * * *

RSA ratio and RR Interval

GDRR SDRR RSA ratio RR interval assist off assist on

distance height start period finish period climbing period side view 237 507

• 6.3˚

3.4˚ 5.7˚

• 2.5˚

[m]

SRR LRR assist off assist on

Control of Torque-Assisted Bicycle based on Physical Activity during Repetitive Prolonged Cycling Exercise, ISEK2004 in Boston, June 19, 2004

• T. Kiryu, Niigata University

RSA_ratio at finish period

RSA ratio before & after the Steep Uphill

distance height start period finish period climbing period side view 237 507

• 6.3˚

3.4˚ 5.7˚

• 2.5˚

[m]

RSA_ratio at start period RSA_ratio at start period RSA_ratio at climbing period

10 20 30 40 50 10 20 30 40 50 10 20 30 40 50 10 20 30 40 50

[%] [%] [%] [%]

SDRR GDRR SDRR GDRR climbing steep uphill

before after

Control of Torque-Assisted Bicycle based on Physical Activity during Repetitive Prolonged Cycling Exercise, ISEK2004 in Boston, June 19, 2004

• T. Kiryu, Niigata University

0.2 0.4 0.6 0.8 0.2 0.4 0.6 0.8

[ s e c ] [ s e c ]

0.2 0.4 0.6 0.8 0.2 0.4 0.6 0.8

[ s e c ] [ s e c ]

RR Interval before & after the Steep Uphill

RRI at finish period

distance height start period finish period climbing period side view 237 507

• 6.3˚

3.4˚ 5.7˚

• 2.5˚

[m]

RR interval at start period RR interval at start period RRI at climbing period LRR SRR LRR SRR climbing steep uphill

before after

Control of Torque-Assisted Bicycle based on Physical Activity during Repetitive Prolonged Cycling Exercise, ISEK2004 in Boston, June 19, 2004

• T. Kiryu, Niigata University

Speed and Torque

5 10 15 20 25 phases 1 and 2 phases 3 and 4 phases 5 and 6

[km/h]

p < 0.05

* * * * * *

5 10 15 20 25 30 35 phases 1 and 2 phases 3 and 4 phases 5 and 6

[Nm]

p < 0.05

* * * * distance height start period finish period climbing period side view 237 507

• 6.3˚

3.4˚ 5.7˚

• 2.5˚

[m]

GDRR SDRR assist off assist on SRR LRR torque speed

Control of Torque-Assisted Bicycle based on Physical Activity during Repetitive Prolonged Cycling Exercise, ISEK2004 in Boston, June 19, 2004

• T. Kiryu, Niigata University

ARV-MPF and ARV-trq

distance height start period finish period climbing period side view 237 507

• 6.3˚

3.4˚ 5.7˚

• 2.5˚

[m]

GDRR SDRR assist off assist on SRR LRR ARV-trq ARV-MPF

Control of Torque-Assisted Bicycle based on Physical Activity during Repetitive Prolonged Cycling Exercise, ISEK2004 in Boston, June 19, 2004

• T. Kiryu, Niigata University

Torque and RSA ratio

ARV-trq at climbing period torque at climbing period RSA_ratio at finish period

distance height start period finish period climbing period side view 237 507

• 6.3˚

3.4˚ 5.7˚

• 2.5˚

[m]

RSA_ratio at finish period

10 20 30 40 50 10 20 30 40 50

SDRR GDRR climbing steep uphill and after

• 1
• 0.8
• 0.6
• 0.4
• 0.2

0.2 0.4 0.6 0.8 1 10 20 30 40 50

GDRR SDRR

after after

[Nm] [%] [%]

Control of Torque-Assisted Bicycle based on Physical Activity during Repetitive Prolonged Cycling Exercise, ISEK2004 in Boston, June 19, 2004

• T. Kiryu, Niigata University

ARV-trq vs RSA ratio

distance height start period finish period climbing period side view 237 507

• 6.3˚

3.4˚ 5.7˚

• 2.5˚

[m]

• 1
• 0.8
• 0.6
• 0.4
• 0.2

0.2 0.4 0.6 0.8 1 10 20 30 40 50

RSA_ratio at finish period

[%]

SDRR GDRR

after

• 1
• 0.8
• 0.6
• 0.4
• 0.2

0.2 0.4 0.6 0.8 1 10 20 30 40 50

RSA_ratio at climbing period

[%]

SDRR GDRR

climbing

• 1
• 0.8
• 0.6
• 0.4
• 0.2

0.2 0.4 0.6 0.8 1 10 20 30 40 50

ARV-trq at climbing period RSA_ratio at start period

[%]

SDRR GDRR

before

Control of Torque-Assisted Bicycle based on Physical Activity during Repetitive Prolonged Cycling Exercise, ISEK2004 in Boston, June 19, 2004

• T. Kiryu, Niigata University

Muscular Fatigue at Each Stroke

ARV-trq at climbing period

mpf difference immediately before the hilltop

distance height

start period finish period climbing period

Torque [Nm]

FH (first half) SH (second half)

time [sec] STFT result EMG [µV]

[Hz]

J J J J J J J J J J J J J J J J J J J J J J J J J J J J J J J JJ J J J J J J J J J J J J J J J J J J J J J J

• 150
• 100
• 50

50 100 150 0.2 0.4 0.6 0.8 1

GDRR SDRR steep slope

Dividing EMG signal into the first and second half of a contraction based on the reference of peak torque, then estimate the change in MPF at each contraction.

Focusing muscle activity just just before the end of climbing

Control of Torque-Assisted Bicycle based on Physical Activity during Repetitive Prolonged Cycling Exercise, ISEK2004 in Boston, June 19, 2004

• T. Kiryu, Niigata University
• In GDRR, mpf at second half decreased during climbing.
• Samples of SDRR and GDRR overlapped

Change of Muscle Activity in a Stroke

just after the 2nd corner immediately before the hilltop

50 100 150 200 250 300 350 50 100 150 200 250 300 350 50 100 150 200 250 300 350 50 100 150 200 250 300 350

SDRR GDRR

[Hz]

mpf at second half

[Hz]

mpf at second half

[Hz]

mpf at first half

[Hz]

mpf at first half

Degeneration of autonomic nervous activity was not related to muscular fatigue

Control of Torque-Assisted Bicycle based on Physical Activity during Repetitive Prolonged Cycling Exercise, ISEK2004 in Boston, June 19, 2004

• T. Kiryu, Niigata University

Discussion

To reflect physical activity on power assist, information on neuromusculer system and autonomic nervous system is separately required.

• Power-assist support muscle force, but does

not directly support autonomic regulation

Muscular system Cardiovascular system

Supply oxygen Removal metabolic byproducts

GDRR in assist-on trials

• muscle activity based on mpf difference at each stroke

Not all of the assist-on trials became the SDRR state

• Grouping the physical states by RSA-ratio

Degeneration of autonomic nervous activity were not related to muscular fatigue

Control of Torque-Assisted Bicycle based on Physical Activity during Repetitive Prolonged Cycling Exercise, ISEK2004 in Boston, June 19, 2004

• T. Kiryu, Niigata University

Conclusion

• 1. Measuring and analyzing biosignals during cycling with a power-assisted bicycle,
• two states, SDRR and GDRR, were divided by the RSA-ratio, and
• muscle fatigue at each stroke was evaluated in relation to the autonomic regulation.
• 2. From 13 subjects and 103 trials,
• conventional power assist strategy is not suitable for rider’s physical work capacity or

conditions, and

• degeneration of autonomic nervous activity was not related to muscular fatigue.
• 3. New strategy of control based on physical activity
• RSA-ratio at start period should determine a long-term assist control.
• MPF difference at each stroke (practically, at specific strokes) should determine a short-

term assist control.

Next step will be a practical study with installing new control strategy for power-assisted bicycles