2000
Pavan, E; Ferrarin, M; Frigo, C
Characterisation of Moment-Angle Relationships in FES Activated Muscles for the Recovery of Sit-To-Stand Inproceedings
In: Sinkjær, Thomas; Popovic, Dejan B; Struijk, Johannes J (Ed.): pp. 141, IFESS_2000, Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark, 2000, ISBN: 87-90562-01-1.
Abstract | Links | BibTeX | Tags: Control and Modeling for Neuroprostheses
@inproceedings{IFESS2000_141,
title = {Characterisation of Moment-Angle Relationships in FES Activated Muscles for the Recovery of Sit-To-Stand},
author = {E Pavan and M Ferrarin and C Frigo},
editor = {Thomas Sinkjær and Dejan B Popovic and Johannes J Struijk},
url = {https://ifess.org/files/proceedings/IFESS2000/IFESS2000_141_Pavan.pdf},
isbn = {87-90562-01-1},
year = {2000},
date = {2000-07-01},
pages = {141},
publisher = {IFESS_2000},
address = {Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark},
abstract = {A method for the identification of dynamic characteristics of the lower leg muscles in paralysed individuals was developed. The set-up was optimised to obtain the joint moments at different joint angles for different muscle groups during the sit-to-stand (STS) movement. The subject seated on a custom system, which allowed him to statically reproduce the STS movement positions, while surface FES was used to elicit isometric contractions in quadriceps, gluteus and hamstrings. From the ground reaction force, acquired by a force platform, the active torque-output was computed by modelling the biomechanical system. The knee joint moment-angle relationships obtained were validated by measuring the forces evoked using a strain gauge based bench. With the above set-up, the active torque contributions of both mono- and bi-articular muscles, for different combinations of them, was quantified avoiding fatigue effects and in a safe way. This characterisation procedure let torques and activation sequences to be estimated and to be used e.g. for improving FES control strategies.},
keywords = {Control and Modeling for Neuroprostheses},
pubstate = {published},
tppubtype = {inproceedings}
}
A method for the identification of dynamic characteristics of the lower leg muscles in paralysed individuals was developed. The set-up was optimised to obtain the joint moments at different joint angles for different muscle groups during the sit-to-stand (STS) movement. The subject seated on a custom system, which allowed him to statically reproduce the STS movement positions, while surface FES was used to elicit isometric contractions in quadriceps, gluteus and hamstrings. From the ground reaction force, acquired by a force platform, the active torque-output was computed by modelling the biomechanical system. The knee joint moment-angle relationships obtained were validated by measuring the forces evoked using a strain gauge based bench. With the above set-up, the active torque contributions of both mono- and bi-articular muscles, for different combinations of them, was quantified avoiding fatigue effects and in a safe way. This characterisation procedure let torques and activation sequences to be estimated and to be used e.g. for improving FES control strategies.
Brown, I E; Cheng, E; Lan, N; Davoodi, R; Loeb, G E
A Comprehensive Model of Muscle Force Generation Under Dynamic Physiological Conditions Inproceedings
In: Sinkjær, Thomas; Popovic, Dejan B; Struijk, Johannes J (Ed.): pp. 127, IFESS_2000, Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark, 2000, ISBN: 87-90562-01-1.
Abstract | Links | BibTeX | Tags: Control and Modeling for Neuroprostheses
@inproceedings{IFESS2000_127,
title = {A Comprehensive Model of Muscle Force Generation Under Dynamic Physiological Conditions},
author = {I E Brown and E Cheng and N Lan and R Davoodi and G E Loeb},
editor = {Thomas Sinkjær and Dejan B Popovic and Johannes J Struijk},
url = {https://ifess.org/files/proceedings/IFESS2000/IFESS2000_127_Brown.pdf},
isbn = {87-90562-01-1},
year = {2000},
date = {2000-07-01},
pages = {127},
publisher = {IFESS_2000},
address = {Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark},
abstract = {The design of controllers for FES and their fitting to individual patients generally requires a mathematical representation of the input-output properties of the stimulated muscles. The inputs are the strength and frequency of the stimulation pulses and the length and velocity of the musculotendinous path; the outputs are total force generation, including active and passive components. The model presented here incorporates extensive new data about the active and passive force-generating properties of muscles under physiological conditions of length, velocity, firing-rate and potentiation. It is available as a Matlab program with a graphical interface for specific muscle morphometries, which generates Simulink blocks suitable for incorporation into models of the kinetics and control of complete musculoskeletal systems.},
keywords = {Control and Modeling for Neuroprostheses},
pubstate = {published},
tppubtype = {inproceedings}
}
The design of controllers for FES and their fitting to individual patients generally requires a mathematical representation of the input-output properties of the stimulated muscles. The inputs are the strength and frequency of the stimulation pulses and the length and velocity of the musculotendinous path; the outputs are total force generation, including active and passive components. The model presented here incorporates extensive new data about the active and passive force-generating properties of muscles under physiological conditions of length, velocity, firing-rate and potentiation. It is available as a Matlab program with a graphical interface for specific muscle morphometries, which generates Simulink blocks suitable for incorporation into models of the kinetics and control of complete musculoskeletal systems.
Ponikvar, M; Munih, M
Computer System for Ankle Joint Muscle Identification Inproceedings
In: Sinkjær, Thomas; Popovic, Dejan B; Struijk, Johannes J (Ed.): pp. 135, IFESS_2000, Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark, 2000, ISBN: 87-90562-01-1.
Abstract | Links | BibTeX | Tags: Control and Modeling for Neuroprostheses
@inproceedings{IFESS2000_135,
title = {Computer System for Ankle Joint Muscle Identification},
author = {M Ponikvar and M Munih},
editor = {Thomas Sinkjær and Dejan B Popovic and Johannes J Struijk},
url = {https://ifess.org/files/proceedings/IFESS2000/IFESS2000_135_Ponikvar.pdf},
isbn = {87-90562-01-1},
year = {2000},
date = {2000-07-01},
pages = {135},
publisher = {IFESS_2000},
address = {Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark},
abstract = {The poster gives an overview of a complex system designed for functional electrical stimulation (FES) assisted standing experiments. The overall system comprises a personal computer (PC), Optotrak system for non-contact position measuring, two force plates, force-sensor shoe insoles, computer controlled electrical stimulator and mechanical rotating frame (MRF). The software core is based on universal simulation environment Matlab Simulink. When stimulating, the sample times of individual sensor device blocks are synchronized with the stimulation pulses. The system functioning was verified in on-line ankle joint muscle identification trials with intact subject and proved to be accurate and reliable.},
keywords = {Control and Modeling for Neuroprostheses},
pubstate = {published},
tppubtype = {inproceedings}
}
The poster gives an overview of a complex system designed for functional electrical stimulation (FES) assisted standing experiments. The overall system comprises a personal computer (PC), Optotrak system for non-contact position measuring, two force plates, force-sensor shoe insoles, computer controlled electrical stimulator and mechanical rotating frame (MRF). The software core is based on universal simulation environment Matlab Simulink. When stimulating, the sample times of individual sensor device blocks are synchronized with the stimulation pulses. The system functioning was verified in on-line ankle joint muscle identification trials with intact subject and proved to be accurate and reliable.
Hunt, K J; Jaime, R-P; Gollee, H; Donaldson, N. deN.
Control of Ankle Joint Stiffness using FES while Standing Inproceedings
In: Sinkjær, Thomas; Popovic, Dejan B; Struijk, Johannes J (Ed.): pp. 131, IFESS_2000, Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark, 2000, ISBN: 87-90562-01-1.
Abstract | Links | BibTeX | Tags: Control and Modeling for Neuroprostheses
@inproceedings{IFESS2000_131,
title = {Control of Ankle Joint Stiffness using FES while Standing},
author = {K J Hunt and R-P Jaime and H Gollee and N.deN. Donaldson},
editor = {Thomas Sinkjær and Dejan B Popovic and Johannes J Struijk},
url = {https://ifess.org/files/proceedings/IFESS2000/IFESS2000_131_Hunt.pdf},
isbn = {87-90562-01-1},
year = {2000},
date = {2000-07-01},
pages = {131},
publisher = {IFESS_2000},
address = {Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark},
abstract = {We have been developing feedback systems for control of unsupported standing in paraplegia. In the present paper we report results of an investigation of ankle stiffness control via functional electrical stimulation (FES), motivated by the need for sufficient ankle stiffness within unsupported standing schemes which aim to harness voluntary and sta- bilising motor control inputs from the upper body. The work was carried out with intact subjects using apparatus in which the subject stands with all joints above the ankles braced, and where ankle moment is provided via FES of the ankle exor and extensor muscles. The results show: (i) that accurate ankle stiffness control, up to the fundamental strength limits of the muscles, can be achieved with controlled FES; (ii) that ankle stiffness control has the potential to ease the task of stabilising upright posture by application of additional upper-body forces.},
keywords = {Control and Modeling for Neuroprostheses},
pubstate = {published},
tppubtype = {inproceedings}
}
We have been developing feedback systems for control of unsupported standing in paraplegia. In the present paper we report results of an investigation of ankle stiffness control via functional electrical stimulation (FES), motivated by the need for sufficient ankle stiffness within unsupported standing schemes which aim to harness voluntary and sta- bilising motor control inputs from the upper body. The work was carried out with intact subjects using apparatus in which the subject stands with all joints above the ankles braced, and where ankle moment is provided via FES of the ankle exor and extensor muscles. The results show: (i) that accurate ankle stiffness control, up to the fundamental strength limits of the muscles, can be achieved with controlled FES; (ii) that ankle stiffness control has the potential to ease the task of stabilising upright posture by application of additional upper-body forces.
Scott, T R D; Atmore, L; Heasman, J M; Flynn, R Y; Vare, V A; Gschwind, C; Middleton, J W; Rutkowski, S B
Control Systems for Stimulated Pronation with the Stimulated Grasp of Persons with Tetraplegia Inproceedings
In: Sinkjær, Thomas; Popovic, Dejan B; Struijk, Johannes J (Ed.): pp. 132, IFESS_2000, Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark, 2000, ISBN: 87-90562-01-1.
Abstract | Links | BibTeX | Tags: Control and Modeling for Neuroprostheses
@inproceedings{IFESS2000_132,
title = {Control Systems for Stimulated Pronation with the Stimulated Grasp of Persons with Tetraplegia},
author = {T R D Scott and L Atmore and J M Heasman and R Y Flynn and V A Vare and C Gschwind and J W Middleton and S B Rutkowski},
editor = {Thomas Sinkjær and Dejan B Popovic and Johannes J Struijk},
url = {https://ifess.org/files/proceedings/IFESS2000/IFESS2000_132_Scott.pdf},
isbn = {87-90562-01-1},
year = {2000},
date = {2000-07-01},
pages = {132},
publisher = {IFESS_2000},
address = {Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark},
abstract = {The pronated posture is used often in relationship to grasp for the acquisition of objects. Subsequent to this, the arm is often moved, whilst maintaining grasp, into the supinated posture for improved stability of grasp. People with tetraplegia having impaired hand grasp and release usually have impaired ability to pronate their arm. Active supination is usually available to them via flexion of the biceps muscle. Electrical stimulation has been incorporated in this study in the provision of three possible methods of control of pronosupination in concert with stimulated hand grasp. Stimulated pronation has been provided to two subjects with tetraplegia secondary to spinal cord injury for the purpose of augmenting stimulated hand grasp and release. Grasp and release has been provided using fully implanted stimulators and stimulation of the pronator musculature provided using surface stimulation. Initially, the instrumented arm has been held in a custom splint with the hand being free to grasp and the degree of pronation held but adjustable. The ability of stimulated grasp to hold an object at varying degrees of pronosupination has been measured and has verified the dependency of grasp stability on the degree of pronosupination and the increased stability of the supinated posture. Subsequently, the three methods integrating the use of pronation during object acquisition and supination for increased grasp stability were examined. These are based on (i) actively controlled stimulation; (ii) touch controlled stimulation; and (iii) constant stimulation of the pronator musculature. Each of the three control methods were assessed during a standardised test involving object acquisition in the pronated posture followed by stable grasp in the supinated posture and completion time was measured. The touch-controlled method, although perhaps the most technically complex, was the significantly fastest (P < 0.02) method of the three in both subjects.},
keywords = {Control and Modeling for Neuroprostheses},
pubstate = {published},
tppubtype = {inproceedings}
}
The pronated posture is used often in relationship to grasp for the acquisition of objects. Subsequent to this, the arm is often moved, whilst maintaining grasp, into the supinated posture for improved stability of grasp. People with tetraplegia having impaired hand grasp and release usually have impaired ability to pronate their arm. Active supination is usually available to them via flexion of the biceps muscle. Electrical stimulation has been incorporated in this study in the provision of three possible methods of control of pronosupination in concert with stimulated hand grasp. Stimulated pronation has been provided to two subjects with tetraplegia secondary to spinal cord injury for the purpose of augmenting stimulated hand grasp and release. Grasp and release has been provided using fully implanted stimulators and stimulation of the pronator musculature provided using surface stimulation. Initially, the instrumented arm has been held in a custom splint with the hand being free to grasp and the degree of pronation held but adjustable. The ability of stimulated grasp to hold an object at varying degrees of pronosupination has been measured and has verified the dependency of grasp stability on the degree of pronosupination and the increased stability of the supinated posture. Subsequently, the three methods integrating the use of pronation during object acquisition and supination for increased grasp stability were examined. These are based on (i) actively controlled stimulation; (ii) touch controlled stimulation; and (iii) constant stimulation of the pronator musculature. Each of the three control methods were assessed during a standardised test involving object acquisition in the pronated posture followed by stable grasp in the supinated posture and completion time was measured. The touch-controlled method, although perhaps the most technically complex, was the significantly fastest (P < 0.02) method of the three in both subjects.
Mourselas, N; Granat, M H
Correction of Drop Foot Using a Fuzzy Logic Controlled Miniature Stimulator Inproceedings
In: Sinkjær, Thomas; Popovic, Dejan B; Struijk, Johannes J (Ed.): pp. 140, IFESS_2000, Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark, 2000, ISBN: 87-90562-01-1.
Abstract | Links | BibTeX | Tags: Control and Modeling for Neuroprostheses
@inproceedings{IFESS2000_140,
title = {Correction of Drop Foot Using a Fuzzy Logic Controlled Miniature Stimulator},
author = {N Mourselas and M H Granat},
editor = {Thomas Sinkjær and Dejan B Popovic and Johannes J Struijk},
url = {https://ifess.org/files/proceedings/IFESS2000/IFESS2000_140_Mourselas.pdf},
isbn = {87-90562-01-1},
year = {2000},
date = {2000-07-01},
pages = {140},
publisher = {IFESS_2000},
address = {Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark},
abstract = {Purpose: To design, implement and evaluate a miniature stimulator which would incorporate closed loop control of ankle dorsiflexion for the swing phase of gait. The system was to be of such physical dimensions and complexity to enable its use by patients on a daily basis. Methods: A miniature stimulator was built which had a fuzzy logic controller in the embedded processor. The derived closed loop control system used a simple resistive goniometer to monitor ankle flexion angle and a force sensitive resistor to detect heel-ground contact. The microcontroller continuously read these two sensors and adjusted the intensity of the stimulation according to the embedded control law. The system was initially tested in the laboratory on three subjects and was compared to an open loop system. It was subsequently given to two subjects to use outside the laboratory and the performance was evaluated and compared to an open loop system over the course of a day. Results: The fuzzy logic controlled stimulator performed better than the open loop controlled stimulator showing an increase in number of steps for which ground clearance was achieved, a reduction in step to step variability and improvements in the quality of foot ground contact. Conclusion: The closed loop control system was used by patients outside the laboratory in their own environment. This system had consistently better performance than an open loop system both in and outside the laboratory.},
keywords = {Control and Modeling for Neuroprostheses},
pubstate = {published},
tppubtype = {inproceedings}
}
Purpose: To design, implement and evaluate a miniature stimulator which would incorporate closed loop control of ankle dorsiflexion for the swing phase of gait. The system was to be of such physical dimensions and complexity to enable its use by patients on a daily basis. Methods: A miniature stimulator was built which had a fuzzy logic controller in the embedded processor. The derived closed loop control system used a simple resistive goniometer to monitor ankle flexion angle and a force sensitive resistor to detect heel-ground contact. The microcontroller continuously read these two sensors and adjusted the intensity of the stimulation according to the embedded control law. The system was initially tested in the laboratory on three subjects and was compared to an open loop system. It was subsequently given to two subjects to use outside the laboratory and the performance was evaluated and compared to an open loop system over the course of a day. Results: The fuzzy logic controlled stimulator performed better than the open loop controlled stimulator showing an increase in number of steps for which ground clearance was achieved, a reduction in step to step variability and improvements in the quality of foot ground contact. Conclusion: The closed loop control system was used by patients outside the laboratory in their own environment. This system had consistently better performance than an open loop system both in and outside the laboratory.
Norton, J A; Donaldson, N. N. de; Day, B L; Dekker, L; Perkins, T A; Wood, D E; McFadden, C; Tromans, A M
The Determinants of Posture in Paraplegics Standing using Lumbar Anterior Root Stimulation Inproceedings
In: Sinkjær, Thomas; Popovic, Dejan B; Struijk, Johannes J (Ed.): pp. 145, IFESS_2000, Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark, 2000, ISBN: 87-90562-01-1.
Abstract | Links | BibTeX | Tags: Control and Modeling for Neuroprostheses
@inproceedings{IFESS2000_145,
title = {The Determinants of Posture in Paraplegics Standing using Lumbar Anterior Root Stimulation},
author = {J A Norton and N.N.de Donaldson and B L Day and L Dekker and T A Perkins and D E Wood and C McFadden and A M Tromans},
editor = {Thomas Sinkjær and Dejan B Popovic and Johannes J Struijk},
url = {https://ifess.org/files/proceedings/IFESS2000/IFESS2000_145_Norton.pdf},
isbn = {87-90562-01-1},
year = {2000},
date = {2000-07-01},
pages = {145},
publisher = {IFESS_2000},
address = {Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark},
abstract = {We have measured the hip extension moment in a patient standing using anterior nerve root stimulation. We present a novel method for calculating the mass properties of the leg using a 3D laser scanner. The joint moments when standing are not significantly different from those recorded in the recumbent posture. We asked the patient to alter the hip angle and present data that suggests that the poor standing posture of the patient may partly be due to a sharp increase in the flexion moment as the hip extends. Only slight changes in both M-wave and non M-wave activity were seen in a number of muscles. As such we believe that the primary reason for the poor posture seen in this patient is biomechanical rather than neurophysiological in nature.},
keywords = {Control and Modeling for Neuroprostheses},
pubstate = {published},
tppubtype = {inproceedings}
}
We have measured the hip extension moment in a patient standing using anterior nerve root stimulation. We present a novel method for calculating the mass properties of the leg using a 3D laser scanner. The joint moments when standing are not significantly different from those recorded in the recumbent posture. We asked the patient to alter the hip angle and present data that suggests that the poor standing posture of the patient may partly be due to a sharp increase in the flexion moment as the hip extends. Only slight changes in both M-wave and non M-wave activity were seen in a number of muscles. As such we believe that the primary reason for the poor posture seen in this patient is biomechanical rather than neurophysiological in nature.
Lyons, G M; Wilcox, D J; Lyons, D J; Hilton, D
Evaluation of a Drop Foot Stimulator FES Intensity Envelope Matched to Tibialis Anterior Muscle Activity during Walking Inproceedings
In: Sinkjær, Thomas; Popovic, Dejan B; Struijk, Johannes J (Ed.): pp. 130, IFESS_2000, Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark, 2000, ISBN: 87-90562-01-1.
Abstract | Links | BibTeX | Tags: Control and Modeling for Neuroprostheses
@inproceedings{IFESS2000_130,
title = {Evaluation of a Drop Foot Stimulator FES Intensity Envelope Matched to Tibialis Anterior Muscle Activity during Walking},
author = {G M Lyons and D J Wilcox and D J Lyons and D Hilton},
editor = {Thomas Sinkjær and Dejan B Popovic and Johannes J Struijk},
url = {https://ifess.org/files/proceedings/IFESS2000/IFESS2000_130_Lyons.pdf},
isbn = {87-90562-01-1},
year = {2000},
date = {2000-07-01},
pages = {130},
publisher = {IFESS_2000},
address = {Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark},
abstract = {A "natural" stimulation strategy is proposed based on muscle activation patterns observed in healthy gait. This stimulation approach was tested on a 52-year old hemiplegic drop foot subject who is ten years post-stroke using a computer-based FES control system. Dorsiflexion angle range was recorded for the subject while walking without orthosis, walking with the "natural" stimulation approach and walking with the conventional trapezoidal stimulation approach. The "natural" approach was found to result in a 76 % increase in dorsiflexion range for the subject, while using only on average 53 % of the stimulation used by the conventional approach. The improved performance of the "natural" stimulation approach is attributed in this case to the subject's calf spasticity, which was rated on the modified Ashworth scale as 3-4, indicating moderate to severe spasticity. The "natural" stimulation envelope is thought to result in a less severe spastic reaction of the calf muscles during dorsiflexion than the conventional trapezoidal approach, enabling a greater dorsiflexion range as a result.},
keywords = {Control and Modeling for Neuroprostheses},
pubstate = {published},
tppubtype = {inproceedings}
}
A "natural" stimulation strategy is proposed based on muscle activation patterns observed in healthy gait. This stimulation approach was tested on a 52-year old hemiplegic drop foot subject who is ten years post-stroke using a computer-based FES control system. Dorsiflexion angle range was recorded for the subject while walking without orthosis, walking with the "natural" stimulation approach and walking with the conventional trapezoidal stimulation approach. The "natural" approach was found to result in a 76 % increase in dorsiflexion range for the subject, while using only on average 53 % of the stimulation used by the conventional approach. The improved performance of the "natural" stimulation approach is attributed in this case to the subject's calf spasticity, which was rated on the modified Ashworth scale as 3-4, indicating moderate to severe spasticity. The "natural" stimulation envelope is thought to result in a less severe spastic reaction of the calf muscles during dorsiflexion than the conventional trapezoidal approach, enabling a greater dorsiflexion range as a result.
Santa-Cruz, M C; Riso, R R; Sepulveda, F
Evaluation of Neural Network Parameters Towards Enhanced Recognition of Naturally Evoked EMG for Prosthetic Hand Grasp Control Inproceedings
In: Sinkjær, Thomas; Popovic, Dejan B; Struijk, Johannes J (Ed.): pp. 137, IFESS_2000, Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark, 2000, ISBN: 87-90562-01-1.
Abstract | Links | BibTeX | Tags: Control and Modeling for Neuroprostheses
@inproceedings{IFESS2000_137,
title = {Evaluation of Neural Network Parameters Towards Enhanced Recognition of Naturally Evoked EMG for Prosthetic Hand Grasp Control},
author = {M C Santa-Cruz and R R Riso and F Sepulveda},
editor = {Thomas Sinkjær and Dejan B Popovic and Johannes J Struijk},
url = {https://ifess.org/files/proceedings/IFESS2000/IFESS2000_137_Santa-Cruz.pdf},
isbn = {87-90562-01-1},
year = {2000},
date = {2000-07-01},
pages = {137},
publisher = {IFESS_2000},
address = {Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark},
abstract = {Myoelectric prosthesis function can be enhanced with different grasp modalities such as palmar and lateral grasp. In the present investigation we recorded percutaneously from four residual muscles (flexor dig; ext. dig; flex. pollicis longus; ext. pollicis longus) of a below elbow (BE) amputee female subject while she contracted her residual muscles according to several computer animations representing the different grasps. Artificial Neural Network (ANN) techniques were applied to discriminate between the 4 intended movements (closing and opening of palmar grasp, and closing and opening of lateral grasp). In the present paper we compared ANNs with different characteristics to determine which combination would enhance performance and reliability. We evaluated variations in two ANN parameters; transfer function and different normalization boundaries for the selected EMG feature. The selected EMG feature was the Mean Absolute Value calculated from the EMG of each residual muscle. We applied the Correct Classification Measure coefficient developed by Triolo et al. [7] to compare the results from the ANNs. The best results were obtained with the following characteristics: sigmoidal transfer function in the output neurons and normalization boundaries selected as the maximum and minimum achieved across different experimental days plus a security margin applied to the upper boundary.},
keywords = {Control and Modeling for Neuroprostheses},
pubstate = {published},
tppubtype = {inproceedings}
}
Myoelectric prosthesis function can be enhanced with different grasp modalities such as palmar and lateral grasp. In the present investigation we recorded percutaneously from four residual muscles (flexor dig; ext. dig; flex. pollicis longus; ext. pollicis longus) of a below elbow (BE) amputee female subject while she contracted her residual muscles according to several computer animations representing the different grasps. Artificial Neural Network (ANN) techniques were applied to discriminate between the 4 intended movements (closing and opening of palmar grasp, and closing and opening of lateral grasp). In the present paper we compared ANNs with different characteristics to determine which combination would enhance performance and reliability. We evaluated variations in two ANN parameters; transfer function and different normalization boundaries for the selected EMG feature. The selected EMG feature was the Mean Absolute Value calculated from the EMG of each residual muscle. We applied the Correct Classification Measure coefficient developed by Triolo et al. [7] to compare the results from the ANNs. The best results were obtained with the following characteristics: sigmoidal transfer function in the output neurons and normalization boundaries selected as the maximum and minimum achieved across different experimental days plus a security margin applied to the upper boundary.
Freschi, C; Vecchi, F; Micera, S; Sabatini, A M; Dario, P
Force Control during Grasp using FES Techniques: Preliminary Results Inproceedings
In: Sinkjær, Thomas; Popovic, Dejan B; Struijk, Johannes J (Ed.): pp. 150, IFESS_2000, Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark, 2000, ISBN: 87-90562-01-1.
Abstract | Links | BibTeX | Tags: Control and Modeling for Neuroprostheses
@inproceedings{IFESS2000_150,
title = {Force Control during Grasp using FES Techniques: Preliminary Results},
author = {C Freschi and F Vecchi and S Micera and A M Sabatini and P Dario},
editor = {Thomas Sinkjær and Dejan B Popovic and Johannes J Struijk},
url = {https://ifess.org/files/proceedings/IFESS2000/IFESS2000_150_Freschi.pdf},
isbn = {87-90562-01-1},
year = {2000},
date = {2000-07-01},
pages = {150},
publisher = {IFESS_2000},
address = {Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark},
abstract = {This work concerns the design and the implementation of feedback controllers to regulate the properties of electrically stimulated muscle. Two different closed-loop controllers ( a PID and a fuzzy controller) have been developed to regulate the force produced during grasp by modulating the stimulation parameters in real-time. The controllers were implemented using a C++ program running on a PC. In each experiment, the recruitment characteristics of the different muscles were first obtained in order to provoke the flexion of the fingers and the thumb. Then, the stimulation parameters of one muscle (the Flexor Digitorum Superficialis) were closed-loop controlled to provoke the hand closing/opening in order to obtain the desired value of the force chosen by the subject. The results indicate that the fuzzy algorithm seems to be more able to deal with the complexity of the controlled system.},
keywords = {Control and Modeling for Neuroprostheses},
pubstate = {published},
tppubtype = {inproceedings}
}
This work concerns the design and the implementation of feedback controllers to regulate the properties of electrically stimulated muscle. Two different closed-loop controllers ( a PID and a fuzzy controller) have been developed to regulate the force produced during grasp by modulating the stimulation parameters in real-time. The controllers were implemented using a C++ program running on a PC. In each experiment, the recruitment characteristics of the different muscles were first obtained in order to provoke the flexion of the fingers and the thumb. Then, the stimulation parameters of one muscle (the Flexor Digitorum Superficialis) were closed-loop controlled to provoke the hand closing/opening in order to obtain the desired value of the force chosen by the subject. The results indicate that the fuzzy algorithm seems to be more able to deal with the complexity of the controlled system.
Cikajlo, I; Bajd, T
Haptic Interface for FES Orthosis Inproceedings
In: Sinkjær, Thomas; Popovic, Dejan B; Struijk, Johannes J (Ed.): pp. 133, IFESS_2000, Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark, 2000, ISBN: 87-90562-01-1.
Abstract | Links | BibTeX | Tags: Control and Modeling for Neuroprostheses
@inproceedings{IFESS2000_133,
title = {Haptic Interface for FES Orthosis},
author = {I Cikajlo and T Bajd},
editor = {Thomas Sinkjær and Dejan B Popovic and Johannes J Struijk},
url = {https://ifess.org/files/proceedings/IFESS2000/IFESS2000_133_Cikajlo.pdf},
isbn = {87-90562-01-1},
year = {2000},
date = {2000-07-01},
pages = {133},
publisher = {IFESS_2000},
address = {Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark},
abstract = {One-degree of freedom haptic interface is proposed for peroneal electrical stimulator. Single channel of functional electrical stimulation (FES) provokes ankle dorsiflexion during walking. FES amplitude is manually controlled by a lever built into the crutch handle. The measured ankle joint angle provides the feedback information, presented to the user as a force feedback applied to the control lever. As the first step in the development of a complex micromechatronic device, a simulated testing environment was provided. A computer model, comprising dynamic ankle foot characteristics, as well as properties of the agonistic and antagonistic muscle groups, substitutes the ankle joint. The model includes fatiguing of the electrically stimulated muscles. A group of six healthy persons was involved in testing of the proposed haptic interface. It was demonstrated that fatiguing of ankle dorsiflexors, stimulated by intermittent pulse trains, can be efficiently overcome by the voluntary control of the peroneal brace through the haptic interface.},
keywords = {Control and Modeling for Neuroprostheses},
pubstate = {published},
tppubtype = {inproceedings}
}
One-degree of freedom haptic interface is proposed for peroneal electrical stimulator. Single channel of functional electrical stimulation (FES) provokes ankle dorsiflexion during walking. FES amplitude is manually controlled by a lever built into the crutch handle. The measured ankle joint angle provides the feedback information, presented to the user as a force feedback applied to the control lever. As the first step in the development of a complex micromechatronic device, a simulated testing environment was provided. A computer model, comprising dynamic ankle foot characteristics, as well as properties of the agonistic and antagonistic muscle groups, substitutes the ankle joint. The model includes fatiguing of the electrically stimulated muscles. A group of six healthy persons was involved in testing of the proposed haptic interface. It was demonstrated that fatiguing of ankle dorsiflexors, stimulated by intermittent pulse trains, can be efficiently overcome by the voluntary control of the peroneal brace through the haptic interface.
Hansen, M; Haugland, M K; Kostov, A; Sinkjaer, T
Machine Learning for Real Time Control of Foot-Drop Correction using Natural Sensors Inproceedings
In: Sinkjær, Thomas; Popovic, Dejan B; Struijk, Johannes J (Ed.): pp. 138, IFESS_2000, Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark, 2000, ISBN: 87-90562-01-1.
Abstract | Links | BibTeX | Tags: Control and Modeling for Neuroprostheses
@inproceedings{IFESS2000_138,
title = {Machine Learning for Real Time Control of Foot-Drop Correction using Natural Sensors},
author = {M Hansen and M K Haugland and A Kostov and T Sinkjaer},
editor = {Thomas Sinkjær and Dejan B Popovic and Johannes J Struijk},
url = {https://ifess.org/files/proceedings/IFESS2000/IFESS2000_138_Hansen.pdf},
isbn = {87-90562-01-1},
year = {2000},
date = {2000-07-01},
pages = {138},
publisher = {IFESS_2000},
address = {Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark},
abstract = {The neural signal recorded from the Sural nerve of a 32 year old female participant (five year post stroke with a foot drop) was used to drive a functional electrical stimulation (FES) system for correction of foot drop. The gait events used to control the stimulation patterns (heel strike and foot lift-off) were detected using a machine learning technique operated from a personal computer. The stimulation was applied to the Common Peroneal nerve using an implanted two-channel stimulator and the electroneurogram (ENG) was recorded using an implantable neural amplifier. Both of the implanted devices were operated using telemetry links. The quality of the detection was evaluated during plain gait, pausing, and stair climbing. A functional performance measure of the detection accuracy approached 100% during plain gait, while in other tasks it was still better than 70%.},
keywords = {Control and Modeling for Neuroprostheses},
pubstate = {published},
tppubtype = {inproceedings}
}
The neural signal recorded from the Sural nerve of a 32 year old female participant (five year post stroke with a foot drop) was used to drive a functional electrical stimulation (FES) system for correction of foot drop. The gait events used to control the stimulation patterns (heel strike and foot lift-off) were detected using a machine learning technique operated from a personal computer. The stimulation was applied to the Common Peroneal nerve using an implanted two-channel stimulator and the electroneurogram (ENG) was recorded using an implantable neural amplifier. Both of the implanted devices were operated using telemetry links. The quality of the detection was evaluated during plain gait, pausing, and stair climbing. A functional performance measure of the detection accuracy approached 100% during plain gait, while in other tasks it was still better than 70%.
Riener, R; Fuhr, T; Antico, D; Quintern, J
Modelling and Control of the Flexion Withdrawal Reflex for Gait Neuroprostheses Inproceedings
In: Sinkjær, Thomas; Popovic, Dejan B; Struijk, Johannes J (Ed.): pp. 128, IFESS_2000, Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark, 2000, ISBN: 87-90562-01-1.
Abstract | Links | BibTeX | Tags: Control and Modeling for Neuroprostheses
@inproceedings{IFESS2000_128,
title = {Modelling and Control of the Flexion Withdrawal Reflex for Gait Neuroprostheses},
author = {R Riener and T Fuhr and D Antico and J Quintern},
editor = {Thomas Sinkjær and Dejan B Popovic and Johannes J Struijk},
url = {https://ifess.org/files/proceedings/IFESS2000/IFESS2000_128_Riener.pdf},
isbn = {87-90562-01-1},
year = {2000},
date = {2000-07-01},
pages = {128},
publisher = {IFESS_2000},
address = {Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark},
abstract = {The flexion withdrawal reflex can be used to initiate the swing phase of gait in Functional Electrical Stimulation (FES). The purpose of this study is to model the input-output relationship of the reflex, thus, providing the basis for a model-based control design. Experimental data were available that describe the reflex response (ankle, knee, and hip joint angles) for stimulation trains with varying input parameters pulse width, frequency, and train duration. An inverse dynamic approach was applied to determine the joint moments from the recorded joint motions. For each joint a set of four characteristic values of the moment pattern (latency, rise time, duration, maximum moment) could be extracted in order to describe the relation between the computed joint moments and the stimulation input. A fuzzy logic approach was used to map any arbitrary stimulation input to the characteristic values. On the basis of these characteristic values the underlying joint moment was reconstructed by a triangular moment pattern. Finally, a physiological model including 3 DoF equations of motions yield the joint movement as model output. Considerably good agreement was obtained between experimental and simulated reflex responses. The model will be used for the design of a swing phase controller in gait neuroprostheses.},
keywords = {Control and Modeling for Neuroprostheses},
pubstate = {published},
tppubtype = {inproceedings}
}
The flexion withdrawal reflex can be used to initiate the swing phase of gait in Functional Electrical Stimulation (FES). The purpose of this study is to model the input-output relationship of the reflex, thus, providing the basis for a model-based control design. Experimental data were available that describe the reflex response (ankle, knee, and hip joint angles) for stimulation trains with varying input parameters pulse width, frequency, and train duration. An inverse dynamic approach was applied to determine the joint moments from the recorded joint motions. For each joint a set of four characteristic values of the moment pattern (latency, rise time, duration, maximum moment) could be extracted in order to describe the relation between the computed joint moments and the stimulation input. A fuzzy logic approach was used to map any arbitrary stimulation input to the characteristic values. On the basis of these characteristic values the underlying joint moment was reconstructed by a triangular moment pattern. Finally, a physiological model including 3 DoF equations of motions yield the joint movement as model output. Considerably good agreement was obtained between experimental and simulated reflex responses. The model will be used for the design of a swing phase controller in gait neuroprostheses.
Obreza, P; Benko, H; Kuzelicki, J; Bajd, T; Kamnik, R; Stefancic, M
Multichannel FES Standing-Up Inproceedings
In: Sinkjær, Thomas; Popovic, Dejan B; Struijk, Johannes J (Ed.): pp. 143, IFESS_2000, Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark, 2000, ISBN: 87-90562-01-1.
Abstract | Links | BibTeX | Tags: Control and Modeling for Neuroprostheses
@inproceedings{IFESS2000_143,
title = {Multichannel FES Standing-Up},
author = {P Obreza and H Benko and J Kuzelicki and T Bajd and R Kamnik and M Stefancic},
editor = {Thomas Sinkjær and Dejan B Popovic and Johannes J Struijk},
url = {https://ifess.org/files/proceedings/IFESS2000/IFESS2000_143_Obreza.pdf},
isbn = {87-90562-01-1},
year = {2000},
date = {2000-07-01},
pages = {143},
publisher = {IFESS_2000},
address = {Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark},
abstract = {Spinal cord injuried (SCI) subjects can raise from a chair by the help of arm support and open-loop electrical stimulation of the knee extensors. The knee joint torques produced by electrically stimulated knee extensors are normally low. Thus extensive effort of upper extremities is required. The aim of the investigation was to study the influence of the stimulation of ankle plantar flexors and hip extensors on the efficiency of standing-up process. SCI subject performed raising from sitting to standing position with surface electrical stimulation of different combinations of muscle groups. The stimulation sequences depended on the phase of the sit-to-stand process. The current phase was detected by measuring the vertical handle reaction force. A robot wrist sensor was used to asses the handle reaction force. The stimulation amplitude for each particular muscle group was scaled to fit between the treshold and saturation value of the recruitment curve. The body kinematics was assessed with contactless optical system. The ground and seat reaction forces were measured with force plates. A 3-dimensional model was used to compute the joint torques. No major differences were noticed between the different types of raising. Additional electrodes on hip extensors and ankle plantar flexors seem not to be justified for daily use of functional electrical stimulation. Standing up is a realtively quick process. It appears that the patient cannot react properly in order to use his stimulated lower extremities to support the standing-up to a larger extent.},
keywords = {Control and Modeling for Neuroprostheses},
pubstate = {published},
tppubtype = {inproceedings}
}
Spinal cord injuried (SCI) subjects can raise from a chair by the help of arm support and open-loop electrical stimulation of the knee extensors. The knee joint torques produced by electrically stimulated knee extensors are normally low. Thus extensive effort of upper extremities is required. The aim of the investigation was to study the influence of the stimulation of ankle plantar flexors and hip extensors on the efficiency of standing-up process. SCI subject performed raising from sitting to standing position with surface electrical stimulation of different combinations of muscle groups. The stimulation sequences depended on the phase of the sit-to-stand process. The current phase was detected by measuring the vertical handle reaction force. A robot wrist sensor was used to asses the handle reaction force. The stimulation amplitude for each particular muscle group was scaled to fit between the treshold and saturation value of the recruitment curve. The body kinematics was assessed with contactless optical system. The ground and seat reaction forces were measured with force plates. A 3-dimensional model was used to compute the joint torques. No major differences were noticed between the different types of raising. Additional electrodes on hip extensors and ankle plantar flexors seem not to be justified for daily use of functional electrical stimulation. Standing up is a realtively quick process. It appears that the patient cannot react properly in order to use his stimulated lower extremities to support the standing-up to a larger extent.
Erfanian, A; Crago, P E
Neural Network Modeling of Electrically Stimulated Muscle under Non-Isometric Conditions Inproceedings
In: Sinkjær, Thomas; Popovic, Dejan B; Struijk, Johannes J (Ed.): pp. 129, IFESS_2000, Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark, 2000, ISBN: 87-90562-01-1.
Abstract | Links | BibTeX | Tags: Control and Modeling for Neuroprostheses
@inproceedings{IFESS2000_129,
title = {Neural Network Modeling of Electrically Stimulated Muscle under Non-Isometric Conditions},
author = {A Erfanian and P E Crago},
editor = {Thomas Sinkjær and Dejan B Popovic and Johannes J Struijk},
url = {https://ifess.org/files/proceedings/IFESS2000/IFESS2000_129_Erfanian.pdf},
isbn = {87-90562-01-1},
year = {2000},
date = {2000-07-01},
pages = {129},
publisher = {IFESS_2000},
address = {Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark},
abstract = {3/4This paper is concerned with developing a force-generating model of electrically stimulated muscle under non-isometric condition based upon artificial neural networks. The stochastic stimulation patterns were applied to the sciatic nerve that innervates the soleus muscle of adult cats using a monopolar spiral cuff electrode. During stimulation, the muscle length is perturbed randomly about a nominal length. We employed the multilayer perceptron (MLP) with back-propagation learning algorithm and Radial Basis Function (RBF) network with stochastic gradient learning rule for muscle modeling, where the stimulation signal, muscle length, velocity of length perturbation, and past measured or predicted force constitute the input of the neural model and the predicted force is the output. The results show that time-varying neural networks would be able to track the muscle force with prediction error 4.1% after the convergence. In addition, when muscle force measurements are not available, the muscle force can be predicted using a time-invariant neural network with mean prediction error 12.7%, 15.2%, and 16.7% for muscle length perturbations of 6 mm, 3 mm, and 1.5 mm, respectively.},
keywords = {Control and Modeling for Neuroprostheses},
pubstate = {published},
tppubtype = {inproceedings}
}
3/4This paper is concerned with developing a force-generating model of electrically stimulated muscle under non-isometric condition based upon artificial neural networks. The stochastic stimulation patterns were applied to the sciatic nerve that innervates the soleus muscle of adult cats using a monopolar spiral cuff electrode. During stimulation, the muscle length is perturbed randomly about a nominal length. We employed the multilayer perceptron (MLP) with back-propagation learning algorithm and Radial Basis Function (RBF) network with stochastic gradient learning rule for muscle modeling, where the stimulation signal, muscle length, velocity of length perturbation, and past measured or predicted force constitute the input of the neural model and the predicted force is the output. The results show that time-varying neural networks would be able to track the muscle force with prediction error 4.1% after the convergence. In addition, when muscle force measurements are not available, the muscle force can be predicted using a time-invariant neural network with mean prediction error 12.7%, 15.2%, and 16.7% for muscle length perturbations of 6 mm, 3 mm, and 1.5 mm, respectively.
Nielsen, K D; Struijk, J J; Sinkjaer, T
New five year Biomedical Engineering curriculum at Aalborg University Inproceedings
In: Sinkjær, Thomas; Popovic, Dejan B; Struijk, Johannes J (Ed.): pp. 151, IFESS_2000, Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark, 2000, ISBN: 87-90562-01-1.
Abstract | Links | BibTeX | Tags: Control and Modeling for Neuroprostheses
@inproceedings{IFESS2000_151,
title = {New five year Biomedical Engineering curriculum at Aalborg University},
author = {K D Nielsen and J J Struijk and T Sinkjaer},
editor = {Thomas Sinkjær and Dejan B Popovic and Johannes J Struijk},
url = {https://ifess.org/files/proceedings/IFESS2000/IFESS2000_151_Nielsen.pdf},
isbn = {87-90562-01-1},
year = {2000},
date = {2000-07-01},
pages = {151},
publisher = {IFESS_2000},
address = {Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark},
abstract = {A new five year curriculum in Biomedical Engineering (MSc) has been established at Aalborg University. The curriculum reflects the multidisciplinary composition of Biomedical Engineering and it contains elements from the medical, engineering and natural sciences as well as humanities and the social sciences. The main objectives are to focus on the abilities and competencies that are required for biomedical engineers; this curriculum intends to deliver engineers with a good background in mathematics and physics and who are thoroughly trained in the engineering systems-approach. The curriculum follows the AAU study form with problem-oriented and project-organized studies. The first five semesters provides basic training in biomedical engineering, mathematics, physics, chemistry, physiology, electronics and applied computer science. In the last five semester three specialities are provided: Sensors, signals and systems (SSS), Medical Informatics (MI) and Bio-mechanical engineering (BM). The curriculum starts September 1st 2000.},
keywords = {Control and Modeling for Neuroprostheses},
pubstate = {published},
tppubtype = {inproceedings}
}
A new five year curriculum in Biomedical Engineering (MSc) has been established at Aalborg University. The curriculum reflects the multidisciplinary composition of Biomedical Engineering and it contains elements from the medical, engineering and natural sciences as well as humanities and the social sciences. The main objectives are to focus on the abilities and competencies that are required for biomedical engineers; this curriculum intends to deliver engineers with a good background in mathematics and physics and who are thoroughly trained in the engineering systems-approach. The curriculum follows the AAU study form with problem-oriented and project-organized studies. The first five semesters provides basic training in biomedical engineering, mathematics, physics, chemistry, physiology, electronics and applied computer science. In the last five semester three specialities are provided: Sensors, signals and systems (SSS), Medical Informatics (MI) and Bio-mechanical engineering (BM). The curriculum starts September 1st 2000.
Schauer, T; Hunt, K J
Nonlinear Predictive Control of Knee-Joint Angle using FES Inproceedings
In: Sinkjær, Thomas; Popovic, Dejan B; Struijk, Johannes J (Ed.): pp. 134, IFESS_2000, Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark, 2000, ISBN: 87-90562-01-1.
Abstract | Links | BibTeX | Tags: Control and Modeling for Neuroprostheses
@inproceedings{IFESS2000_134,
title = {Nonlinear Predictive Control of Knee-Joint Angle using FES},
author = {T Schauer and K J Hunt},
editor = {Thomas Sinkjær and Dejan B Popovic and Johannes J Struijk},
url = {https://ifess.org/files/proceedings/IFESS2000/IFESS2000_134_Schauer.pdf},
isbn = {87-90562-01-1},
year = {2000},
date = {2000-07-01},
pages = {134},
publisher = {IFESS_2000},
address = {Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark},
abstract = {Nonlinear model-based predictive control (NMPC) based on a recurrent neural network (RNN) is applied to tracking control of the knee-joint angle using functional electrical stimulation (FES) of the quadriceps muscle group. The feedback control system was first investigated using a complex biomechanical model of the human knee of a sitting subject. The nonlinear predictive controller was compared with a linear discrete pole-placement controller which is determined using an ARX model of the knee-joint dynamics at one operating point. The simulation results show that nonlinear predictive control is superior to linear control with respect to the achievable tracking performance over the entire operating range. The RNN has proved to be suitable for modelling the inputoutput behaviour of the knee joint dynamics.},
keywords = {Control and Modeling for Neuroprostheses},
pubstate = {published},
tppubtype = {inproceedings}
}
Nonlinear model-based predictive control (NMPC) based on a recurrent neural network (RNN) is applied to tracking control of the knee-joint angle using functional electrical stimulation (FES) of the quadriceps muscle group. The feedback control system was first investigated using a complex biomechanical model of the human knee of a sitting subject. The nonlinear predictive controller was compared with a linear discrete pole-placement controller which is determined using an ARX model of the knee-joint dynamics at one operating point. The simulation results show that nonlinear predictive control is superior to linear control with respect to the achievable tracking performance over the entire operating range. The RNN has proved to be suitable for modelling the inputoutput behaviour of the knee joint dynamics.
Mann, G E; Burridge, J H; Ewins, D J; McLellan, D L; Swain, I D; Taylor, P N; Wood, D E; Wright, P A
Optimising Two Channel Stimulation to Improve Walking Following Stroke Inproceedings
In: Sinkjær, Thomas; Popovic, Dejan B; Struijk, Johannes J (Ed.): pp. 139, IFESS_2000, Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark, 2000, ISBN: 87-90562-01-1.
Abstract | Links | BibTeX | Tags: Control and Modeling for Neuroprostheses
@inproceedings{IFESS2000_139,
title = {Optimising Two Channel Stimulation to Improve Walking Following Stroke},
author = {G E Mann and J H Burridge and D J Ewins and D L McLellan and I D Swain and P N Taylor and D E Wood and P A Wright},
editor = {Thomas Sinkjær and Dejan B Popovic and Johannes J Struijk},
url = {https://ifess.org/files/proceedings/IFESS2000/IFESS2000_139_Mann.pdf},
isbn = {87-90562-01-1},
year = {2000},
date = {2000-07-01},
pages = {139},
publisher = {IFESS_2000},
address = {Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark},
abstract = {The aims of the study were: To identify patients most likely to benefit from two channel lower limb stimulation with the microcontroller based Compustim 10B stimulator. To investigate whether optimum muscle groups and control algorithms can be selected by simple clinical assessment or whether specialist gait analysis is required. To investigate the orthotic and re-education effects of using a second channel of stimulation. To develop a library of control algorithms for walking 14 subjects (mean age 63.2, sd 6.7 years, 9 male 5 female) who had been treated with the single channel Odstock Dropped Foot Stimulator (ODFS) for a minimum of 6 months were recruited. All subjects followed a Control - Treatment - Control (ABA) study design where Treatment was two channel stimulation with the Compustim 10B and Control was single channel common peroneal stimulation with the Compustim 10B. Walking speed and Physiological Cost Index were measured at approximately four - weekly intervals throughout the study. Gait analysis was performed at weeks 0, 12, 24 and 36. Results indicate a significant therapeutic and orthotic effect on walking speed from using a second channel of stimulation. There was poor correlation between clinical observation of gait and specialist gait analysis for both predicted algorithms and evaluation of correction of identified gait problems. Both clinical observation of gait and specialist gait analysis have their limitations. It may be more relevant to use a scoring system to analyse gait with a simple tool to collect ankle and knee angle data, rather than expensive specialist equipment. Video analysis to aid clinical observation would also improve accuracy.},
keywords = {Control and Modeling for Neuroprostheses},
pubstate = {published},
tppubtype = {inproceedings}
}
The aims of the study were: To identify patients most likely to benefit from two channel lower limb stimulation with the microcontroller based Compustim 10B stimulator. To investigate whether optimum muscle groups and control algorithms can be selected by simple clinical assessment or whether specialist gait analysis is required. To investigate the orthotic and re-education effects of using a second channel of stimulation. To develop a library of control algorithms for walking 14 subjects (mean age 63.2, sd 6.7 years, 9 male 5 female) who had been treated with the single channel Odstock Dropped Foot Stimulator (ODFS) for a minimum of 6 months were recruited. All subjects followed a Control - Treatment - Control (ABA) study design where Treatment was two channel stimulation with the Compustim 10B and Control was single channel common peroneal stimulation with the Compustim 10B. Walking speed and Physiological Cost Index were measured at approximately four - weekly intervals throughout the study. Gait analysis was performed at weeks 0, 12, 24 and 36. Results indicate a significant therapeutic and orthotic effect on walking speed from using a second channel of stimulation. There was poor correlation between clinical observation of gait and specialist gait analysis for both predicted algorithms and evaluation of correction of identified gait problems. Both clinical observation of gait and specialist gait analysis have their limitations. It may be more relevant to use a scoring system to analyse gait with a simple tool to collect ankle and knee angle data, rather than expensive specialist equipment. Video analysis to aid clinical observation would also improve accuracy.
Thorsen, R; Veltink, P; Ferrarin, M; Frigo, C
Precision of Myo-electrically Controlled Functional Electrical Stimulation of the Paretic Hand after a Spinal Cord Lesion Inproceedings
In: Sinkjær, Thomas; Popovic, Dejan B; Struijk, Johannes J (Ed.): pp. 149, IFESS_2000, Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark, 2000, ISBN: 87-90562-01-1.
Abstract | Links | BibTeX | Tags: Control and Modeling for Neuroprostheses
@inproceedings{IFESS2000_149,
title = {Precision of Myo-electrically Controlled Functional Electrical Stimulation of the Paretic Hand after a Spinal Cord Lesion},
author = {R Thorsen and P Veltink and M Ferrarin and C Frigo},
editor = {Thomas Sinkjær and Dejan B Popovic and Johannes J Struijk},
url = {https://ifess.org/files/proceedings/IFESS2000/IFESS2000_149_Thorsen.pdf},
isbn = {87-90562-01-1},
year = {2000},
date = {2000-07-01},
pages = {149},
publisher = {IFESS_2000},
address = {Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark},
abstract = {A Myoelectric Controlled Functional Electrical Stimulator (MeCFES) for restoring the hand function has been tested by 6 patients with spinal cord lesion. We have utilised residual myo-electric signal from the wrist extensor (m. extensor carpi radialis, ECR) to control functional electrical stimulation (FES) of either wrist extension (i.e. the same muscle) or thumb flexion. The accuracy of the resulting movement with and without the MeCFES was evaluated by use of a tracking test, where the user was trying to follow a target representing the exerted force. Wrist extension force was improved in 2 of 5 C5-spinal cord injured (SCI) tetraplegics. Wrist controlled thumb flexion was tested by one C3 incomplete SCI person with significant increase of isometric force. From the tracking tests, the tendency seems to be that low residual voluntary force results in less accuracy of the movement but with significant increase of the muscle output. This paper shows experiments concerned with the use of the myoelectric signal (MES) for control of FES.},
keywords = {Control and Modeling for Neuroprostheses},
pubstate = {published},
tppubtype = {inproceedings}
}
A Myoelectric Controlled Functional Electrical Stimulator (MeCFES) for restoring the hand function has been tested by 6 patients with spinal cord lesion. We have utilised residual myo-electric signal from the wrist extensor (m. extensor carpi radialis, ECR) to control functional electrical stimulation (FES) of either wrist extension (i.e. the same muscle) or thumb flexion. The accuracy of the resulting movement with and without the MeCFES was evaluated by use of a tracking test, where the user was trying to follow a target representing the exerted force. Wrist extension force was improved in 2 of 5 C5-spinal cord injured (SCI) tetraplegics. Wrist controlled thumb flexion was tested by one C3 incomplete SCI person with significant increase of isometric force. From the tracking tests, the tendency seems to be that low residual voluntary force results in less accuracy of the movement but with significant increase of the muscle output. This paper shows experiments concerned with the use of the myoelectric signal (MES) for control of FES.
Kamnik, R; Bajd, T
Robot Assisted FES Standing-Up of Paraplegic Patient Inproceedings
In: Sinkjær, Thomas; Popovic, Dejan B; Struijk, Johannes J (Ed.): pp. 144, IFESS_2000, Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark, 2000, ISBN: 87-90562-01-1.
Abstract | Links | BibTeX | Tags: Control and Modeling for Neuroprostheses
@inproceedings{IFESS2000_144,
title = {Robot Assisted FES Standing-Up of Paraplegic Patient},
author = {R Kamnik and T Bajd},
editor = {Thomas Sinkjær and Dejan B Popovic and Johannes J Struijk},
url = {none},
isbn = {87-90562-01-1},
year = {2000},
date = {2000-07-01},
pages = {144},
publisher = {IFESS_2000},
address = {Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark},
abstract = {none},
keywords = {Control and Modeling for Neuroprostheses},
pubstate = {published},
tppubtype = {inproceedings}
}
none
Sepulveda, F
In: Sinkjær, Thomas; Popovic, Dejan B; Struijk, Johannes J (Ed.): pp. 136, IFESS_2000, Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark, 2000, ISBN: 87-90562-01-1.
Abstract | Links | BibTeX | Tags: Control and Modeling for Neuroprostheses
@inproceedings{IFESS2000_136,
title = {Simulating FES Gait: Radial Basis Function Networks vs. Neuro-Fuzzy Inference and Recurrent Neural Networks with Plant Wear Factors},
author = {F Sepulveda},
editor = {Thomas Sinkjær and Dejan B Popovic and Johannes J Struijk},
url = {https://ifess.org/files/proceedings/IFESS2000/IFESS2000_136_Sepulveda.pdf},
isbn = {87-90562-01-1},
year = {2000},
date = {2000-07-01},
pages = {136},
publisher = {IFESS_2000},
address = {Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark},
abstract = {Introduction: Radial Basis Function Networks (RBFNs) and an Adaptive Network-Based Fuzzy Inference System (ANFIS) were used for simulating the neuromuscular system during FES-induced gait swing. These methods were compared to previous research with recurrent neural networks that included indexes related to muscle fatigue and reflex habituation. Real clinical data were used both for training and for evaluating the various models. Methods: Two RBFN structures were developed for simulating FES-based gait. The simplest structure was to map frame by frame input-output relationships, while for the other structure each input-output vector pair included angular data from one entire gait swing cycle. The model's inputs were pulse width values applied to the common peroneal and femoral nerves, respectively, while the outputs were predicted knee and ankle flexion/extension angles. Also, two independent Sugeno-type ANFIS models were trained for making frame-by-frame predictions of knee and ankle angles, respectively. Results and Discussion: The worst results were obtained with the RBFN structure based on gait cycle to gait cycle predictions (as opposed to a structure dealing with intra-cycle, frame by frame predictions). The results for both RBFNs were also significantly worse than for the previously studied recurrent networks with wear factors. Poor performance (both in training and in testing modes) was also obtained with the ANFIS model. These observations differ from those of other groups, which may stem from the fact that in this study real, noisy data were used both for training and for testing the various models. The superiority of the recurrent networks probably comes from their ability to map dynamic data structures. Thus, it is suggested that data representative of input delays be used with the RBFN and ANFIS structures in the future.},
keywords = {Control and Modeling for Neuroprostheses},
pubstate = {published},
tppubtype = {inproceedings}
}
Introduction: Radial Basis Function Networks (RBFNs) and an Adaptive Network-Based Fuzzy Inference System (ANFIS) were used for simulating the neuromuscular system during FES-induced gait swing. These methods were compared to previous research with recurrent neural networks that included indexes related to muscle fatigue and reflex habituation. Real clinical data were used both for training and for evaluating the various models. Methods: Two RBFN structures were developed for simulating FES-based gait. The simplest structure was to map frame by frame input-output relationships, while for the other structure each input-output vector pair included angular data from one entire gait swing cycle. The model's inputs were pulse width values applied to the common peroneal and femoral nerves, respectively, while the outputs were predicted knee and ankle flexion/extension angles. Also, two independent Sugeno-type ANFIS models were trained for making frame-by-frame predictions of knee and ankle angles, respectively. Results and Discussion: The worst results were obtained with the RBFN structure based on gait cycle to gait cycle predictions (as opposed to a structure dealing with intra-cycle, frame by frame predictions). The results for both RBFNs were also significantly worse than for the previously studied recurrent networks with wear factors. Poor performance (both in training and in testing modes) was also obtained with the ANFIS model. These observations differ from those of other groups, which may stem from the fact that in this study real, noisy data were used both for training and for testing the various models. The superiority of the recurrent networks probably comes from their ability to map dynamic data structures. Thus, it is suggested that data representative of input delays be used with the RBFN and ANFIS structures in the future.
Popovic, M; Popovic, D
Synergistic Control for an Elbow Neuroprosthesis Inproceedings
In: Sinkjær, Thomas; Popovic, Dejan B; Struijk, Johannes J (Ed.): pp. 148, IFESS_2000, Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark, 2000, ISBN: 87-90562-01-1.
Abstract | Links | BibTeX | Tags: Control and Modeling for Neuroprostheses
@inproceedings{IFESS2000_148,
title = {Synergistic Control for an Elbow Neuroprosthesis},
author = {M Popovic and D Popovic},
editor = {Thomas Sinkjær and Dejan B Popovic and Johannes J Struijk},
url = {https://ifess.org/files/proceedings/IFESS2000/IFESS2000_148_Popovic.pdf},
isbn = {87-90562-01-1},
year = {2000},
date = {2000-07-01},
pages = {148},
publisher = {IFESS_2000},
address = {Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark},
abstract = {The proposed control method for the elbow neuroprostheses (NP) assumes that the most proximal non-paralyzed segment drives automatically the paralyzed segments of the upper-limb; in this case, the upper arm controls the forearm. The control uses production rules methodology: the rule-base comprises synergies formed automatically by inductive learning (IL). The IL automatically builds a decision tree based on many examples (kinematic data) and supervised learning. The examples selected for the learning have been recorded in able-bodied subjects while performing typical activities of daily living (eating with a fork, drinking from a can, writing, using a telephone receiver, and pouring from a container). The IL captures the synergy and expresses it in the form of IF-THEN rules: IF the shoulder flexion/extension velocity is given in the three consecutive times, THEN the elbow flexion/extension velocity is predicted. These rules are used for the coordination, being an interface between the volitional and actuator control levels. The actuator level is customized to the individual biomechanical and physiologic properties of the eventual user. This system can be part of both an implantable and/or a surface electrode NP. Initial tests with tetraplegic humans take place with the surface electrodes NP, using the Belgrade grasping system [13]. The results show that subjects can reach objects with required accuracy.},
keywords = {Control and Modeling for Neuroprostheses},
pubstate = {published},
tppubtype = {inproceedings}
}
The proposed control method for the elbow neuroprostheses (NP) assumes that the most proximal non-paralyzed segment drives automatically the paralyzed segments of the upper-limb; in this case, the upper arm controls the forearm. The control uses production rules methodology: the rule-base comprises synergies formed automatically by inductive learning (IL). The IL automatically builds a decision tree based on many examples (kinematic data) and supervised learning. The examples selected for the learning have been recorded in able-bodied subjects while performing typical activities of daily living (eating with a fork, drinking from a can, writing, using a telephone receiver, and pouring from a container). The IL captures the synergy and expresses it in the form of IF-THEN rules: IF the shoulder flexion/extension velocity is given in the three consecutive times, THEN the elbow flexion/extension velocity is predicted. These rules are used for the coordination, being an interface between the volitional and actuator control levels. The actuator level is customized to the individual biomechanical and physiologic properties of the eventual user. This system can be part of both an implantable and/or a surface electrode NP. Initial tests with tetraplegic humans take place with the surface electrodes NP, using the Belgrade grasping system [13]. The results show that subjects can reach objects with required accuracy.
Andrews, B J; Thrasher, A; Chan, A; Davoodi, R
Trunk Momentum can Reduce Upper Limb Forces in FES Aided Paraplegic Locomotion Inproceedings
In: Sinkjær, Thomas; Popovic, Dejan B; Struijk, Johannes J (Ed.): pp. 146, IFESS_2000, Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark, 2000, ISBN: 87-90562-01-1.
Abstract | Links | BibTeX | Tags: Control and Modeling for Neuroprostheses
@inproceedings{IFESS2000_146,
title = {Trunk Momentum can Reduce Upper Limb Forces in FES Aided Paraplegic Locomotion},
author = {B J Andrews and A Thrasher and A Chan and R Davoodi},
editor = {Thomas Sinkjær and Dejan B Popovic and Johannes J Struijk},
url = {https://ifess.org/files/proceedings/IFESS2000/IFESS2000_146_Andrews.pdf},
isbn = {87-90562-01-1},
year = {2000},
date = {2000-07-01},
pages = {146},
publisher = {IFESS_2000},
address = {Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark},
abstract = {The use of trunk momentum to assist FES aided paraplegic locomotion was examined via computer simulation and clinical trials. While standing in a pair of medially linked knee-ankle-foot orthoses, paraplegic subject (T12) was able to lift his feet in turn by oscillating his trunk. He was able to do so at frequencies ranging from 0.38 to 0.55 Hz, comparable to swing frequencies of passive gait. An able-bodied subject also performed these movements, and was able to walk at speeds up to 0.61 m/s. Two computer models demonstrated rocking as well, but stable walking patterns could not be found. Another model demonstrated that the Coriolis effect of trunk momentum did not contribute significantly to sit-to-stand transfers.},
keywords = {Control and Modeling for Neuroprostheses},
pubstate = {published},
tppubtype = {inproceedings}
}
The use of trunk momentum to assist FES aided paraplegic locomotion was examined via computer simulation and clinical trials. While standing in a pair of medially linked knee-ankle-foot orthoses, paraplegic subject (T12) was able to lift his feet in turn by oscillating his trunk. He was able to do so at frequencies ranging from 0.38 to 0.55 Hz, comparable to swing frequencies of passive gait. An able-bodied subject also performed these movements, and was able to walk at speeds up to 0.61 m/s. Two computer models demonstrated rocking as well, but stable walking patterns could not be found. Another model demonstrated that the Coriolis effect of trunk momentum did not contribute significantly to sit-to-stand transfers.
Bidard, C; Rienstra, S; Veltink, P H; Koopman, H F J M; Grady, J; deJ. Vries,; Huttenhuis, L
Trunk Stability while Standing or Sitting: A Static Analysis Inproceedings
In: Sinkjær, Thomas; Popovic, Dejan B; Struijk, Johannes J (Ed.): pp. 142, IFESS_2000, Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark, 2000, ISBN: 87-90562-01-1.
Abstract | Links | BibTeX | Tags: Control and Modeling for Neuroprostheses
@inproceedings{IFESS2000_142,
title = {Trunk Stability while Standing or Sitting: A Static Analysis},
author = {C Bidard and S Rienstra and P H Veltink and H F J M Koopman and J Grady and deJ. Vries and L Huttenhuis},
editor = {Thomas Sinkjær and Dejan B Popovic and Johannes J Struijk},
url = {https://ifess.org/files/proceedings/IFESS2000/IFESS2000_142_Bidard.pdf},
isbn = {87-90562-01-1},
year = {2000},
date = {2000-07-01},
pages = {142},
publisher = {IFESS_2000},
address = {Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark},
abstract = {Several neuromuscular disorders cause trunk unstability. External support such as seating aids, orthoses, or external muscle control by FES, are then needed. In this paper, we investigate the static requirement for stabilizing the trunk in the sagittal plane while standing or sitting. A simple biomechanical model is used, composed of two moving segments, the pelvis and the upper trunk. We evaluate the stabilizing effort by the critical stiffness, the minimal linear stiffness required for making the posture stable against the gravity forces. Three basic situations are compared: standing erect, sitting erect unsupported and with a backrest. This preliminary study shows the importance of pelvis posture and motion in stabilizing the trunk. It confirms the importance of pelvis stabilization.},
keywords = {Control and Modeling for Neuroprostheses},
pubstate = {published},
tppubtype = {inproceedings}
}
Several neuromuscular disorders cause trunk unstability. External support such as seating aids, orthoses, or external muscle control by FES, are then needed. In this paper, we investigate the static requirement for stabilizing the trunk in the sagittal plane while standing or sitting. A simple biomechanical model is used, composed of two moving segments, the pelvis and the upper trunk. We evaluate the stabilizing effort by the critical stiffness, the minimal linear stiffness required for making the posture stable against the gravity forces. Three basic situations are compared: standing erect, sitting erect unsupported and with a backrest. This preliminary study shows the importance of pelvis posture and motion in stabilizing the trunk. It confirms the importance of pelvis stabilization.
Gharooni, S; Tokhi, M O; Heller, B
The use of Elastic Element in a Hybrid Orthosis for Swing Phase Generation in Orthotic Gait Inproceedings
In: Sinkjær, Thomas; Popovic, Dejan B; Struijk, Johannes J (Ed.): pp. 147, IFESS_2000, Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark, 2000, ISBN: 87-90562-01-1.
Abstract | Links | BibTeX | Tags: Control and Modeling for Neuroprostheses
@inproceedings{IFESS2000_147,
title = {The use of Elastic Element in a Hybrid Orthosis for Swing Phase Generation in Orthotic Gait},
author = {S Gharooni and M O Tokhi and B Heller},
editor = {Thomas Sinkjær and Dejan B Popovic and Johannes J Struijk},
url = {https://ifess.org/files/proceedings/IFESS2000/IFESS2000_147_Gharooni.pdf},
isbn = {87-90562-01-1},
year = {2000},
date = {2000-07-01},
pages = {147},
publisher = {IFESS_2000},
address = {Center for Sensory-Motor Interaction (SMI), Aalborg University, Denmark},
abstract = {A new technique for generating the swing phase in a passive orthosis is introduced in this paper. A spring is used for knee flexion. The gravity and inertial forces assist hip flexion and artificially stimulated muscle is used as the power source. This spring knee orthosis provides more-natural swing phase trajectory than that produced by the flexion reflex for gait in spinal cord injured subjects. To circumvent the complexity involved in accurately modelling stimulated muscle, a fuzzy logic controller (FLC) was adopted for prediction and for the control scheme. It is shown that hip flexion can be produced without the need for the withdrawal reflex, hip flexor stimulus or any mechanical actuator at the hip. A hip flexion angle of 21 degrees was achieved with a non-impaired subject wearing a prototype orthosis.},
keywords = {Control and Modeling for Neuroprostheses},
pubstate = {published},
tppubtype = {inproceedings}
}
A new technique for generating the swing phase in a passive orthosis is introduced in this paper. A spring is used for knee flexion. The gravity and inertial forces assist hip flexion and artificially stimulated muscle is used as the power source. This spring knee orthosis provides more-natural swing phase trajectory than that produced by the flexion reflex for gait in spinal cord injured subjects. To circumvent the complexity involved in accurately modelling stimulated muscle, a fuzzy logic controller (FLC) was adopted for prediction and for the control scheme. It is shown that hip flexion can be produced without the need for the withdrawal reflex, hip flexor stimulus or any mechanical actuator at the hip. A hip flexion angle of 21 degrees was achieved with a non-impaired subject wearing a prototype orthosis.