study of legs' sensorymotor asymmetry using a lower-limb rehabilitation robot
TRANSCRIPT
Study of leg’s sensorimotor asymmetry using alower-limb rehabilitation robot :
The LegoPress
Chouaieb Nemri
https://www.linkedin.com/in/nemri
Department of Electrical and Control Engineering Robotic Systems LaboratoryENSEEIHT EPFL
September, 2014
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 1 / 34
Outline
1 IntroductionThe LegoPressFlexCOM
2 Study of sensorimotor asymmetryPrevious studiesStudy of lower-limb sensorimotor asymmetry
3 Control design and implementationLeg weight compensationPosition MatchingElastic MatchingEffort Matching
4 ResultsPreliminary observations
5 Conclusion
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 2 / 34
Introduction The LegoPress
Outline
1 IntroductionThe LegoPressFlexCOM
2 Study of sensorimotor asymmetryPrevious studiesStudy of lower-limb sensorimotor asymmetry
3 Control design and implementationLeg weight compensationPosition MatchingElastic MatchingEffort Matching
4 ResultsPreliminary observations
5 Conclusion
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 3 / 34
Introduction The LegoPress
Overview
Simple and cost-effective lower-limb rehabilitation device.
Figure: CAD view of the legopress
Adjustable sitting position - 2 separate motorized axes.
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 4 / 34
Introduction The LegoPress
Axis elements
Figure: CAD view of one axis
Figure: schematic view of the axis elements
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 5 / 34
Introduction The LegoPress
Force transducer
Precise force monitoring can be done by means of force sensorspositioned at the pedal level.
Figure: CAD view of the force sensor
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 6 / 34
Introduction FlexCOM
Outline
1 IntroductionThe LegoPressFlexCOM
2 Study of sensorimotor asymmetryPrevious studiesStudy of lower-limb sensorimotor asymmetry
3 Control design and implementationLeg weight compensationPosition MatchingElastic MatchingEffort Matching
4 ResultsPreliminary observations
5 Conclusion
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 7 / 34
Introduction FlexCOM
PC-based motion control solution
PC-based open architecture with minimum time development
Adapted to different type of robots: serial (cartesian or not) orparallel.
Microsoft windows : Standard development software - standard PCIinput/output.
Real time capabilities guaranteed with RTX extension.
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 8 / 34
Introduction FlexCOM
PC-based motion control solution
PC-based open architecture with minimum time development
Adapted to different type of robots: serial (cartesian or not) orparallel.
Microsoft windows : Standard development software - standard PCIinput/output.
Real time capabilities guaranteed with RTX extension.
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 8 / 34
Introduction FlexCOM
PC-based motion control solution
PC-based open architecture with minimum time development
Adapted to different type of robots: serial (cartesian or not) orparallel.
Microsoft windows : Standard development software - standard PCIinput/output.
Real time capabilities guaranteed with RTX extension.
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 8 / 34
Introduction FlexCOM
PC-based motion control solution
PC-based open architecture with minimum time development
Adapted to different type of robots: serial (cartesian or not) orparallel.
Microsoft windows : Standard development software - standard PCIinput/output.
Real time capabilities guaranteed with RTX extension.
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 8 / 34
Study of sensorimotor asymmetry Previous studies
Outline
1 IntroductionThe LegoPressFlexCOM
2 Study of sensorimotor asymmetryPrevious studiesStudy of lower-limb sensorimotor asymmetry
3 Control design and implementationLeg weight compensationPosition MatchingElastic MatchingEffort Matching
4 ResultsPreliminary observations
5 Conclusion
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 9 / 34
Study of sensorimotor asymmetry Previous studies
Previous studies
Movement, an essential way we have to interact with the world.
Sensorimotor asymmetry has been shown in upper limbs for righthanded people.
Findings are strongly supported by the following dual-linear model
Figure: Dual-linear model of sensori-motor feedback loops of each limb
Quantifiying asymmetries in the lower limbs.
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 10 / 34
Study of sensorimotor asymmetry Previous studies
Previous studies
Movement, an essential way we have to interact with the world.Sensorimotor asymmetry has been shown in upper limbs for righthanded people.
Findings are strongly supported by the following dual-linear model
Figure: Dual-linear model of sensori-motor feedback loops of each limb
Quantifiying asymmetries in the lower limbs.
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 10 / 34
Study of sensorimotor asymmetry Previous studies
Previous studies
Movement, an essential way we have to interact with the world.Sensorimotor asymmetry has been shown in upper limbs for righthanded people.Findings are strongly supported by the following dual-linear model
Figure: Dual-linear model of sensori-motor feedback loops of each limb
Quantifiying asymmetries in the lower limbs.
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 10 / 34
Study of sensorimotor asymmetry Previous studies
Previous studies
Movement, an essential way we have to interact with the world.Sensorimotor asymmetry has been shown in upper limbs for righthanded people.Findings are strongly supported by the following dual-linear model
Figure: Dual-linear model of sensori-motor feedback loops of each limb
Quantifiying asymmetries in the lower limbs.
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 10 / 34
Study of sensorimotor asymmetry Study of lower-limb sensorimotor asymmetry
Outline
1 IntroductionThe LegoPressFlexCOM
2 Study of sensorimotor asymmetryPrevious studiesStudy of lower-limb sensorimotor asymmetry
3 Control design and implementationLeg weight compensationPosition MatchingElastic MatchingEffort Matching
4 ResultsPreliminary observations
5 Conclusion
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 11 / 34
Study of sensorimotor asymmetry Study of lower-limb sensorimotor asymmetry
Objectives
The study aims to assess and understand lower-limb asymmetriessensorimotor origins.
To this end, an experimental protocol featuring postion and/or effortmatching tasks has been designed.
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 12 / 34
Study of sensorimotor asymmetry Study of lower-limb sensorimotor asymmetry
Objectives
The study aims to assess and understand lower-limb asymmetriessensorimotor origins.
To this end, an experimental protocol featuring postion and/or effortmatching tasks has been designed.
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 12 / 34
Study of sensorimotor asymmetry Study of lower-limb sensorimotor asymmetry
Experimental set up
The LegoPress
Reference position : 90 degree knee flexion
Target position : 120 degree knee flexion
Target force : 20% of the maximum voluntary exertion (MVE)
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 13 / 34
Study of sensorimotor asymmetry Study of lower-limb sensorimotor asymmetry
Experimental set up
The LegoPress
Reference position : 90 degree knee flexion
Target position : 120 degree knee flexion
Target force : 20% of the maximum voluntary exertion (MVE)
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 13 / 34
Study of sensorimotor asymmetry Study of lower-limb sensorimotor asymmetry
Experimental set up
The LegoPress
Reference position : 90 degree knee flexion
Target position : 120 degree knee flexion
Target force : 20% of the maximum voluntary exertion (MVE)
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 13 / 34
Study of sensorimotor asymmetry Study of lower-limb sensorimotor asymmetry
Experimental set up
The LegoPress
Reference position : 90 degree knee flexion
Target position : 120 degree knee flexion
Target force : 20% of the maximum voluntary exertion (MVE)
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 13 / 34
Study of sensorimotor asymmetry Study of lower-limb sensorimotor asymmetry
Experimental tasks and conditions
Position and/or effort reference is performed by the reference leg andthence, participants are asked to perform matching with the same oropposite leg.
Different Matching conditions
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 14 / 34
Study of sensorimotor asymmetry Study of lower-limb sensorimotor asymmetry
Experimental tasks and conditions
Position and/or effort reference is performed by the reference leg andthence, participants are asked to perform matching with the same oropposite leg.
Different Matching conditions
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 14 / 34
Study of sensorimotor asymmetry Study of lower-limb sensorimotor asymmetry
Experimental tasks and conditions
18 possible combinations
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 15 / 34
Control design and implementation Leg weight compensation
Outline
1 IntroductionThe LegoPressFlexCOM
2 Study of sensorimotor asymmetryPrevious studiesStudy of lower-limb sensorimotor asymmetry
3 Control design and implementationLeg weight compensationPosition MatchingElastic MatchingEffort Matching
4 ResultsPreliminary observations
5 Conclusion
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 16 / 34
Control design and implementation Leg weight compensation
Leg weight compensation
The horizontal effort is measured every 20 mm along our experimentsuseful portion of the axis range (10 mm to 290 mm).
4th Polynomial regression has been used to identify the model
Figure: 2nd degree polynomial regression is insufficient
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 17 / 34
Control design and implementation Leg weight compensation
Leg weight compensation
The horizontal effort is measured every 20 mm along our experimentsuseful portion of the axis range (10 mm to 290 mm).
4th Polynomial regression has been used to identify the model
Figure: 2nd degree polynomial regression is insufficient
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 17 / 34
Control design and implementation Position Matching
Outline
1 IntroductionThe LegoPressFlexCOM
2 Study of sensorimotor asymmetryPrevious studiesStudy of lower-limb sensorimotor asymmetry
3 Control design and implementationLeg weight compensationPosition MatchingElastic MatchingEffort Matching
4 ResultsPreliminary observations
5 Conclusion
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 18 / 34
Control design and implementation Position Matching
Position matchingAdmittance control
Admittance control : Pedals move proportionally to the activelyapplied force magnitude.
Figure: Block diagram of the admittance controller
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 19 / 34
Control design and implementation Position Matching
Position matchingAdmittance control
A deadzone has been introduced to overcome model’s imperfections(threshold = 10% of passive forces)
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 20 / 34
Control design and implementation Position Matching
Position matchingAdmittance control
The controller laws can be descibed as follows:
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 21 / 34
Control design and implementation Position Matching
Position matching algorithmAdmittance control
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 22 / 34
Control design and implementation Elastic Matching
Outline
1 IntroductionThe LegoPressFlexCOM
2 Study of sensorimotor asymmetryPrevious studiesStudy of lower-limb sensorimotor asymmetry
3 Control design and implementationLeg weight compensationPosition MatchingElastic MatchingEffort Matching
4 ResultsPreliminary observations
5 Conclusion
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 23 / 34
Control design and implementation Elastic Matching
Choice of control paradigm
Proportional position controller
Unaccurate restoring force
Admittance based spring behaviour
Unstable behaviour for low stiffnesses
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 24 / 34
Control design and implementation Elastic Matching
Impedance based spring
Block diagram of the controller
The controller equations are the following
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 25 / 34
Control design and implementation Elastic Matching
Elastic matching algorithm
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 26 / 34
Control design and implementation Effort Matching
Outline
1 IntroductionThe LegoPressFlexCOM
2 Study of sensorimotor asymmetryPrevious studiesStudy of lower-limb sensorimotor asymmetry
3 Control design and implementationLeg weight compensationPosition MatchingElastic MatchingEffort Matching
4 ResultsPreliminary observations
5 Conclusion
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 27 / 34
Control design and implementation Effort Matching
Effort matching
During effort matching tasks, the pedals are simply position-controlled
The pedals are fixed in the 120 degree knee extension position.
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 28 / 34
Control design and implementation Effort Matching
Effort matching
During effort matching tasks, the pedals are simply position-controlled
The pedals are fixed in the 120 degree knee extension position.
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 28 / 34
Control design and implementation Effort Matching
Effort matching algorithm
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 29 / 34
Results Preliminary observations
Outline
1 IntroductionThe LegoPressFlexCOM
2 Study of sensorimotor asymmetryPrevious studiesStudy of lower-limb sensorimotor asymmetry
3 Control design and implementationLeg weight compensationPosition MatchingElastic MatchingEffort Matching
4 ResultsPreliminary observations
5 Conclusion
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 30 / 34
Results Preliminary observations
Preliminary observations
Data has not been fully processed
In ipsilateral condition, position matching is more accurate with theleft leg.
No overshoot (resp. undershoot) trends when the right (resp. left)limb is matching the opposite limb’s reference.
Analysed data is not sufficient enough to make us draw soundconclusions.
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 31 / 34
Results Preliminary observations
Preliminary observations
Data has not been fully processed
In ipsilateral condition, position matching is more accurate with theleft leg.
No overshoot (resp. undershoot) trends when the right (resp. left)limb is matching the opposite limb’s reference.
Analysed data is not sufficient enough to make us draw soundconclusions.
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 31 / 34
Results Preliminary observations
Preliminary observations
Data has not been fully processed
In ipsilateral condition, position matching is more accurate with theleft leg.
No overshoot (resp. undershoot) trends when the right (resp. left)limb is matching the opposite limb’s reference.
Analysed data is not sufficient enough to make us draw soundconclusions.
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 31 / 34
Results Preliminary observations
Preliminary observations
Data has not been fully processed
In ipsilateral condition, position matching is more accurate with theleft leg.
No overshoot (resp. undershoot) trends when the right (resp. left)limb is matching the opposite limb’s reference.
Analysed data is not sufficient enough to make us draw soundconclusions.
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 31 / 34
Conclusion
Conclusion
An experimental protocol featuring position and/or effort matchinghas been designed in order to assess sensorimotor lower-limbasymmetry.
We run the experiments with 6 right-legged and one left-legged youngunimpaired participants.
During position tasks, the matching pedal is admittance controlled.While during elastic matching, the controller behaves as animpedance.
Future work, will focus on processing and thoroughly analysingthecollected data.
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 32 / 34
Appendix References
References I
J. Olivier. M. Jeannaret. M. Bouri. and H. BleulerThe LegoPress : A rehabilitation assessment and performance device -Mechanical Design and Control.LSRO -EPFL adfa, p. 1, 2011. Springer-Verlag Berlin Heidelberg 2011
M. Bouri. and R. ClavelA windows PC-based robot controller - An open architecture.ISR, Tokyo, Japan, November 29 - December 01, 2005
D. .E.Adamo. and B.J. MartinPosition sense asymmetry.Exp Brain Res DOI 10.1007/s00221-008-1560-0
D. .E.Adamo. S.Scotland and B.J. Martinasymmetry in grasp force matching and sense of effort.Exp Brain Res (2012) 217:273-285 DOI 10.1007/s00221-011-2991-6
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 33 / 34
Appendix References
Acknowledgements
Questions?
Chouaieb Nemri (ENSEEIHT – EPFL - LSRO)Study of leg’s sensorimotor asymmetry using a lower-limb rehabilitation robot :September, 2014 34 / 34