bra. di. po and pigro: innovative devices for motor learning

Research ArticleBraDiPO and PIGRO Innovative Devices forMotor Learning Programs

Guido Belforte1 Gabriella Eula1 Silvia Sirolli1 Paolo Bois1 Elisabetta Geda2

Federico DrsquoAgata2 Franco Cauda2 Sergio Duca3 Marina Zettin4 Roberta Virgilio4

Giuliano Geminiani2 and Katiuscia Sacco2

1 Department of Mechanical and Aerospace Engineering Politecnico di Torino Italy2 Department of Psychology Universita di Torino Italy3 Neuroradiology Service Ospedale Koelliker Torino Italy4 Puzzle Social Cooperative Torino Italy

Correspondence should be addressed to Gabriella Eula gabriellaeulapolitoit

Received 14 June 2013 Revised 15 November 2013 Accepted 11 December 2013 Published 11 March 2014

Academic Editor Ryo Saegusa

Copyright copy 2014 Guido Belforte et alThis is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

Two mechatronics prototypes useful for robotic neurotreatments and new clinical trainings are here presented PIGRO(pneumatic interactive gait rehabilitation orthosis) is an active exoskeletonwith an electropneumatic control It imposesmovementson lower limbs in order to produce in the patientrsquos brain proper motor cortex activation BraDiPO (brain discovery pneumaticorthosis) is anMR-compatible device designed to improve fMRI (functionalmagnetic resonance imaging) analysisThe twodevicesare presented together because both are involved in the study of new robotic treatments of patients affected by ictus or brain strokeor in some motor learning experimental investigations carried out on healthy subjects

1 Introduction

According to the theory of neuroplasticity which neurol-ogists have accepted for only a few decades the brain iscapable of ldquolearningrdquo even in adult age and following injuriesif appropriately stimulated [1 2] In recent years a number ofdevices have been developed [3 4] that can stimulate certainfunctions or simulate physiological movements so locomotorand cognitive functions in brain-damaged patients can beinvestigated using fMRI or functional magnetic resonanceimaging [5 6]

Functional magnetic resonance imaging has made itpossible to look into the human brain ldquoin vivordquo for the firsttime literally ldquowatching it at workrdquo In addition fMRI is usedon healthy subjects to gain an understanding of our brainrsquoscomplex capabilities in studies of motor learning

The paper presents two optimised electropneumatic pro-totypes whose previous design can be read in [7ndash12]

The first one called BraDiPO (brain discovery pneu-matic orthosis) is used to move the subjectrsquos feet during the

fMRI analysis in order to impose a controlled movement andstimulate the motor cortex during the test

The other one called PIGRO (pneumatic interactivegait rehabilitation orthosis) is an active electropneumaticexoskeleton for lower limbs motor exercises

Designed at the Politecnico di Torino Department ofMechanical andAerospace Engineering the two devices haveto be used together in the study of motor imagery withhealthy subjects or not to evaluate changes in brain plasticityat the level of motor circuits and motor imagination

Their common andmain advantages are physiotherapistrsquoswork improvement movement imposed with repeatabilityelectronic data acquisition continuous measurement of thewhole test parameters different clinical protocols possibility

In comparison with their previous design [7] theseoptimised prototypes show the improvements underlined inthe following paragraphs

Furthermore the clinical procedure here presented car-ried out using BraDiPO and PIGRO on healthy subjects

Hindawi Publishing CorporationJournal of RoboticsVolume 2014 Article ID 656029 12 pageshttpdxdoiorg1011552014656029

2 Journal of Robotics

is an interesting research on human motor cortex function-ing

2 Robotic Machines Prototypes

21 BraDiPO BraDiPO [7 9ndash11] is an MR-compatibledevice with 1 DoF (degree of freedom) in the sagittal planewhich allows the patientrsquos ankle joint rotation around itsproper axis It has two pedals to which the patientrsquos feetare secured moved by a pneumatic actuator (Figures 1(a)and 1(b)) and it operates either in ldquoactive moderdquo (patientfollows the pedal movement passively) or in ldquopassive moderdquo(patient moves the feet autonomously and an optical angularpotentiometer put on the pedals records the traveled angle)

Some similar examples are shown in literature as dis-cussed in [13ndash23]

These devices are always MR compatible as they haveto work in the magnetic resonance chamber without pro-ducing interference [13ndash15] and use in general pneumatic orhydraulic controls

BraDiPO can be adapted to patientrsquos feet with ananthropometric range between 10ile woman and 95ileman in agreement with the standards underlined in [24 25]All electrical parts and the PC are remotely located A user-friendly graphical interface is also provided

A flow chart of the BraDiPO management software isshown in Figure 1(c) It was developed by the authors inorder to optimize this investigation Each test starts with ageneral system check after which the operator can choosetest parameters (cylinder velocity pedal angle range andoperating frequency) Each frame has a virtual emergencyswitch to stop software during analysis

Themain problems encountered with the first BraDiPOprototype (Figure 1(b)) are referred to the pneumatic actua-tor position improvement the device position adjustment onthe scanning table the dynamics signal transmission outside-inside from the magnetic resonance chamber

Consequently a second prototype was designed as alsodiscussed in [10] In this second solution the pneumaticactuator (1) is located inside the device (Figure 2(a)) and itis connected to a bar (2) which rotates around hinge O (3) tomove the pedals (4)

As it was found during the initial fMRI analysis withBraDiPO the patientrsquos foot movement could interfere withthe test by producing involuntary head movements

So the device was raised as illustrated in Figure 2(b)because this configuration reduces the transmission of move-ment from patientrsquos feet to head and solves the problem

As the raising of the patientrsquos legs also requires a horizon-tal adjustment of the patientrsquos position on the scanning table(Figure 2(c)) another pneumatic actuator was added underthe box and used asmotor of a horizontal positioning system

The final second BraDiPO prototype is shown in Fig-ure 3

Moreover the problem of a proper dynamics signaltransmission through the long tubes here used (about 10meter) widely presented in literature [16ndash23] for other similardevices and by authors in [11] was studied and already solvedin the first BraDiPO prototype

Here the problem was addressed by selecting nonmag-netic commercial valves and placing them close to thecylinder (PF and PD in Figure 4(a))

In particular this study was carried out using commercialsoftware where the layout of the circuit shown in Figure 4(b)was designed

In this way authors investigated the dynamics perfor-mances of BraDiPO control circuit in order to obtain forthe patientrsquos feet movement the required working frequency(about 1Hz) It is imposed by paradigms connected to thehuman motor cortex activation

In this numerical model a number of parameters werechanged in order to analyze the systemrsquos dynamic behaviorvarying pedal loads (m) working frequencies (f ) supplypressures (119901

119904) and duty cycles (d) The latter parameter

is very important because it directly affects the line sup-plydischarge time

During simulation parameters were varied in the fol-lowing ranges mass m = 1-2 kg working frequency f = 02ndash07Hz duty cycle d = 015 025 040 050 and relative supplypressure 119901

119904= 30lowast105 35lowast105 40lowast105 Pa

In order to validate the numerical model elaborated anexperimental test bench was constructed and some tests werecarried out to compare experimental and theoretical results[11]

An example of results obtained from this study is shownin Figure 5 with 2 kg load 35lowast105 Pa supply pressure 05Hzfrequency and 025 duty cycle [11]

These graphs illustrate a good correspondence betweentheoretical and experimental results carried out in the sametest condition In fact varying the various parameters theshape and the amplitude of the signals are always good andshow a proper functioning of the numericalmodel elaboratedby the authors as can be widely checked in [11]

In particular analysing Figure 5 Pc (see also Figure 4(a))reaches the same maximum amplitude value both in thetheoretical and experimental tests Puf (Figure 4(a)) shows atheoretical amplitude value lower than the experimental onePud (Figure 4(a)) shows the same behaviour in both cases

This fact is certainly due to the little errors that sometimesoccur between numerical models and experimental testsThe main causes here considered are cylinder and valvesfrictionmodeling and physical modeling of a long pneumaticline Overall results obtained are good and give possibilityto construct a final prototype and circuit working alwaysproperly

22 PIGRO The other prototype here involved is calledPIGRO and it is an active exoskeleton for lower limbs withan electropneumatic control

The literature on active lower-limb exoskeletons used forgait rehabilitation as discussed in [26ndash31] chiefly deals withmachines tethered to a fixed stationThey are generally ratherrigid heavy devices and are mostly employed for treadmilltrainingThese systems almost never provide ankle actuation

By contrast PIGRO system is an innovative devicedesigned for the rehabilitation of patients with clinical prob-lems such as hemiplegia tetraparesis and hemiparesis aswell

Journal of Robotics 3

(a)

8

16

3

5

2

7

4

(b)

Start

Autodiagnosis

No

Yes

Regulations(angle cylinder velocity)

Yes

No

Tests

Frequency pause

Acquisition dataAcquisition data

Save fileSave file

Emergency

Emergency

Yes

No

No

Yes

Passive mode Active mode

(c)

Figure 1 (a) BraDiPO in the magnetic resonance chamber during test (b) BraDiPO first prototype (c) BraDiPO flow chart


1 2

3 4

A

Box

O

H

120576

(a)

z

yh

120572

(b)

120572120572998400

Δy

Δz

(c)

Figure 2 (a) Schematic viewof themechanismunderlying the second prototype (b)Approach used to determineBraDiPO vertical positionregulation (c) Horizontal movement of the patientrsquos lower limbs as BraDiPO is raised

Figure 3 Second BraDiPO prototype

as those who have suffered strokes ischemias or cerebralhemorrhages It can also be extended to certain cases ofmuscular dystrophy and degenerative motor system disease

Figure 6 shows some details of the systemIn particular the system (Figure 6(a)) consists of an

electropneumatically controlled active orthosis a PC toacquire and process data and enable the operator to managethe session a monitor for the operator and a monitor forpatient biofeedback

In Figure 6(b) an example of screenshot from the opera-torrsquos monitor during the test is shown It is possible to noticethe six joints scopes

In Figures 6(c) 6(d) and 6(e) examples of screenshotsfrom the biofeedback monitor are shown for the patientrsquos

Table 1 PIGRO adjustments

10ilewoman (mm)

95ile man(mm)

Regulationrange (mm)

Pelvis width 300 650 350Femur length 330 500 170Tibia length 330 500 170

hip-knee-ankle In particular the thicker curve is themachinereference for the patient while the thinner curve is thepatientrsquos performance during the test In the biofeedbackmonitor clinicians require a thicker reference curve as rangewithin the patientrsquos performance can be checked during thetreatment

The orthosis is a modular 6-DoF exoskeleton which canbe adapted to patients with anthropometric range between10ile woman and 95ile man [24 25] as shown in Table 1All DoF are in the sagittal plane one for each lower limb joint

Part of the orthosis structure uses spring steel so that thesystem can bemore readily comfortable and wearable as wellas to allow a certain pelvic movement out of the sagittal planeduring the gait cycle

Depending on therapeutic requirements PIGRO canbe used either with the patient suspended or with partialground contact using a body weight support In both casesthe body weight support is used to unload the mass of theorthosis as well as that of the patientrsquos body

As in both types of training the ankle joint actua-tion is essential for the patientrsquos motor cortex activationthis represents an original and important characteristic ofPIGRO [32]


Patient

Magnetic resonance chamber

SE

PE

E A F D

P

OE

R1

R2RE

SE Software emergencyOE Operator emergencyPE Patient emergencyE Emergency supply cutoff valveRE Reset after emergency

A Active mode electrovalveP Passive mode electrovalveF Control plantarflexion electrovalveD Control dorsiflexion electrovalve PF Power plantarflexion pneumatic valve

PD Power dorsiflexion pneumatic valve Rf1 Rf2 Flow resistancesR1 R2 Pressure reducersT1 T2 T3 T4 Pressure transducers

PD

PF

F

Rf1

Rf2

Pc

Puf

T1

T3 T4

T2

Pud

24V

(a)

Rf1

E

F DVc

A

Rf2

PcPF

PuF PuD

PD

Pilot stage PF

Cylinder

Gasdata

Pilot stage PD

CpCp

PPP

CpCpCpCp

A

P

P

T

A

P T

A

P T

A

P

P

T

T AP

TAP

T

M F

k

k k

kkElectric input

+rarr

M+rarrM

X X

+rarr

(b)

Figure 4 (a) Final electropneumatic control circuit (b) Layout of the numerical model simulating the optimized circuit


000510152025303540

000 060 120 180 240 300 360Time (s)

Pc (b

ar)

0

5

10

15

20

25

30

Elec

tric

inpu

t (V

)

Pressure Pc Electric input

00

02

04

06

08

10

12

000 064 128 192 256 320 384Time (s)

Puf

(bar

)

00

02

04

06

08

10

12

Pud

(bar

)

Theoretical pressure Puf

Theoretical pressure Pud

000 060 120 180 240 300 360Time (s)

m = 2kg f = 05Hz d = 05 ps = 35bar

m = 2kg ps = 35bar f = 05Hz d = 05

m = 2kg ps = 35bar f = 05Hz d = 05

(a)

000510152025303540

000 048 096 144 192 240 288 336 384Time (s)

Pc (b

ar)

000 052 104 156 208 260 312 364Time (s)

000 052 104 156 208 260 312 364Time (s)

00

02

04

06

08

10

12

Puf (

bar)

00

02

04

06

08

10

12

Pud

(bar

)

m = 2kg f = 05Hz d = 05 ps = 35bar

m = 2kg f = 05Hz d = 05 ps = 35bar

m = 2kg f = 05Hz d = 05 ps = 35bar

(b)

Figure 5 Theoretical (left) and experimental (right) results with 2 kg load on actuator 35 bar supply pressure 05Hz frequency and 025duty cycle (1 bar = 105 Pa)

In particular the ankle joint actuation can also beremoved leaving the patient free to move the foot indepen-dently during the treatment if required

In particular overground walking was preferred to walk-ing on treadmill as it allows the patientrsquos advancing inthe room and the space provides important sensations andperceptions fundamental to rehabilitation

The pneumatic actuators operate on the principle ofagonistantagonist muscle pair thus reducing weight andbulk The current cylinders could also be replaced withpneumatic muscles if required

Each leg of the orthosis is equipped with cylinderchamber pressure sensors and position sensors which trackjoint movements for use as feedback in system control

Supply pressure level can be regulated both to vary theforce imposed on the patientrsquos legs and to help the physiciansrsquoanalysis of the patientrsquos autonomous walking progress

Actuation is electropneumatic but can also be imple-mented with electric or hydraulic actuators

The management software is a real-time control whereinput curves can be either the physiological joints behaviorof a standard gait cycle [33] or some other curves choice byclinicians for the training

Acquired data are sent to the PC via a 10-meter long codedsignal transmission cable multipolar cables or wirelessconnections The software was developed by the authorsspecifically for this application It also features an excellentgraphical interface to facilitate the operatorrsquos work


(a)

(b) (c)

(d) (e)

Figure 6 (a) PIGRO in a rehabilitation center (b) PIGRO operatorrsquos monitor screenshot with the six joints visualized ((c)(d) and (e))Examples of biofeedback monitor screenshots for hip (c) knee (d) and ankle (e)

This graphical interface allows inserting all patientrsquosparameters especially the anthropometric length of the legand the patientrsquos weight In particular patientrsquos mass caninfluence PIGRO movement both as inertial effects andhuman interaction with the exoskeleton So this value isfundamental for the control system

PIGRO does not exceed 30 kg in weight and it isflexible versatile and easy to use

In particular it must be underlined that this robot is notsuitable for the patientrsquos assistance during the day and outsidethe hospital as other auxiliary devices on the market

In fact PIGRO is a robotic machine designed forneurorehabilitation training carried out in hospital structuresand by clinicians only [32]

Moreover pneumatic actuation is intrinsically safe andclean gives a comfortable and soft imposed movement and


05

10152025

Time (s)

60 66 73 79 86 92 98 105 111 118

Left

hip

angl

e (de

g)

minus5

minus10

minus15

minus20

minus25

InputOutput

(a)

0

10

20

30

40

50

60

70

60 66 73 79 86 92 98 105 111 118

Left

knee

angl

e (de

g)

Time (s)

InputOutput

(b)

Time (s)InputOutput

60 66 73 79 86 92 98 105 111 118

Left

ankl

e ang

le (d

eg)

05

101520

minus5

minus10

minus15

minus20

minus25

minus30

(c)

Figure 7 Comparison between PIGRO input-output joints angles curves in hip (a) knee (b) and ankle (c)

allows changing forces on patientrsquos legs operating on thesupply pressure level

Figures 7(a) 7(b) and 7(c) show PIGRO behaviorcomparing during an experimental test input and outputcurves for each joint angle This test was carried out on ahealthy subject with a weight of about 70 kg In particular thegraphs referred to a gait cycle of 3 s analyzed after the initialtransient state and to the left patientrsquos leg for simplicity

These results underline the proper functioning of thesystem as the amplitude and the shape of PIGRO inputcurves (control position) and output curves (feedback of thesystem) are always in full agreement In particular the smalldelay between input and output curves is due to the fewinteractions that anyway occur between a passive conscioussubject and a robotic imposed movement

Finally the main innovations of this new prototype [32]in comparison with the previous ones [7 8 12] are hereunderlined

An important improvement of the human-machine inter-face design was carried out by authors studying and testinginnovative ergonomic and comfortable textiles structures

A new electric solution with its own separate emergencyswitch was realized for the pelvic adjustment improvingrobot wearability and safety

The control system emergencies were finally defined inthree modes from software with a pneumatic button witha patientrsquos button

An innovative real-time control system was designed inorder to substitute the previous one based on two PC (masterand slave)

PIGRO management software was fully reviewed andimproved

In particular the software is now capable to slowlystart the gait cycle in order to avoid an initial strong andsuddenmovement imposed on the patientrsquos legs to dischargeall electro-pneumatic valves if some emergency situationsoccur or during the pause required in the training to stoppneumatic actuators in pressure when clinicians check thepatientrsquos cognitive status

Furthermore supply pressure for one leg or both can bechanged by the operator during the treatment automaticallysaving in a proper patientrsquos memo block the time of thisaction and the new pressure level

A graphical user-friendly interface was designed in fullagreement with cliniciansrsquo requirements

23 Comparison of the Two Devices Themain characteristicsof the two devices here presented are summarized in Table 2


Table 2 Technical features of the two prototypes

BraDiPO PIGRONonmagnetic materials and controls Versatile hardware and softwareVersatile hardware and software Good wearabilityMinimal invasiveness Minimal invasivenessEasily used emergency controls for operator and patient Applicable to many clinical situationsRemotely located electrical parts Low weightAdjustable to patient body measurements Can also be used in the waterUser-friendly graphical interface Can be used for both suspended and overground walkingFeet can be moved singly or simultaneously in active or passive mode User-friendly graphical interfaceGood wearability Easily used emergency controls for operator and patient

Figure 8 Motor training using PIGRO

Both BraDiPO and PIGRO have a good wearabilityproper and original management software and a usefulgraphical user-friendly interface

They allow repeating exactly each treatment to save datato help and improve the physiotherapistrsquos work

They are versatile and allow testing and establishinginnovative procedure useful for the neurorehabilitation andfor the human brain motor cortex study

The clinical application of these two devices consistsof combination of locomotor and cognitive training herepreliminarily carried out with healthy subjects using fMRIto measure changes in plasticity both at the level of the motorcortex and in motor imagery tasks [34 35]

3 Material and Methods

31 Subjects Five healthy volunteers (Figure 8) (3 womenand 2 men age range = 20ndash23 mean age = 22 years) tookpart in the experiment All subjects were tested and werefound to have a sufficient ability to form visual and motorimages Exclusion criteria included history of neurologicalor developmental illness mental disorders drug or alcoholabuse and current use of medications known to alter neuro-logical activity All subjects gave informed written consentThe fMRI study was performed at the Koelliker Hospital inTorino (Italy)

32 Training Subjects performed the training tasks using arobotic device (PIGRO see below for a description) Thetraining session consisted of two runs Each run includedactive and passive phases During passive phases subjectskept their eyes closed and were asked to accommodatemovements imposed by the robotic device and to focus onkinesthetic perception Movements consisted of a sequenceof ankle dorsi- and plantarflexions movements differed inrhythm and speed for the two feet In the active phase pres-sure in the device decreased and subjects had to reproducethe movements learned in the focusing phase with the sameamplitude and speed Each phase lasted five minutes Thefindings from the motor imagery tasks will be discussedbelow

33 fMRI Assessment fMRI assessment made use of a motorand motor imagery task subjects were required to performankle dorsiflexion and plantarflexion In the second sessionsubjects were required to imagine the same movementComplete dorsiflexionplantarflexion cycles should occurwith a frequency of about 05Hz The task was performedusing BraDiPO Paradigms were performed using a blockdesign with 12 s of rest alternating with 12 s of the activecondition Each paradigm consisted of a total of 25 blocks (13rest conditions 12 active conditions) 4 functional volumeswere scanned during each block each paradigm lasted 5min


Figure 9 Pre- and posttraining differences in the motor imagery task

Table 3 Talairach coordinates of activations in the motor imagery task

119883

coor119884

coor119885

coor 119905 119875 Location

200 10 510 4613314 0000004 Right cerebrum Frontal lobe Subgyral Gray matter Brodmann area 650 610 120 6094687 0000000 Right cerebrum Frontal lobe Medial frontal gyrus Gray matter Brodmann area 10

34 Image Acquisition Data acquisition was performed ona 15 Tesla scanner optimized for functional imaging Func-tional images were acquired using echo planar sequenceswith a repetition time (TR) of 3000ms an echo time (TE)of 60ms and a 90∘ flip angle The acquisition matrix was 64times 64 the field of view (FoV) was 256mm For each paradigma total of 100 volumes were acquired Each volume consistedof 25 axial slices parallel to the anterior-posterior (AC-PC)commissure line and covering the whole brain

4 Results and Discussion

Imaging data were analyzed using the scanner describedabove

After preprocessing a series of steps were performedin order to allow for precise anatomical locations of brainactivity to facilitate intersubject analysis First each subjectrsquosslice-based functional scans were coregistered with their 3Dhigh-resolution structural scan This process involved math-ematical coregistration exploiting slice positioning stored inthe headers of the raw data as well as fine adjustments thatwere computed by comparing the data sets on the basis oftheir intensity values if needed manual adjustments werealso performed Second each subjectrsquos 3D structural data setwas transformed into a Talairach space as discussed in [36]the cerebrum was translated and rotated into the anterior-posterior commissure plane and then the borders of the cere-brumwere identifiedThird using the anatomical-functionalcoregistrationmatrix and the determined Talairach referencepoints each subjectrsquos functional time course was transformedinto a Talairach space and the volume time course wascreated For the motor paradigm the following procedurewas performed A multisubject multistudy design matrixwas specified and each defined box-car was convolved with

a predefined Hemodynamic Response Function (HRF) toaccount for the hemodynamic delay as discussed in [37]A statistical analysis using the general linear model withseparate study predictors was performed on the group toyield functional activationmaps during the pre- and posttestsseparately All voxels activated in the pretest and thoseactivated in the posttest were combined to create a maskexcluding the rest of the cerebrum and cerebellum

This mask was used to compute the general linear modelcomparing posttest activations with pretest activations in thegroup of subjects since the same data set was used for maskdefinition and subsequent statistical tests

A comparison of imaging data obtained before and aftertraining revealed activations in motor imagery task that isposttest increased hemodynamic responses in right frontalgyrus including supplementary motor area and right medialfrontal gyrus (see Figure 9 and Table 3)

Figure 9 illustrates some selected fMRI results showingdifferences before and after training with PIGRO in themotor imagery task

They suggest a cortical reorganization in motor areassuch as the supplementary motor area precuneus andcerebellum after training which may show the effect ofrehabilitation on the reorganization of somatotopic maps

The passive stimulation (BraDiPO in ldquoactive moderdquo)as expected shows a robust sensorimotor supplementarymotor and cerebellar activity plus some temporal andparietalclusters The mean time course shows that the activity in thisarea is strongly correlated with the stimulation paradigm

The active stimulation (BraDiPO in ldquopassive moderdquo) isless affected by head motion noise and shows a less robustsensorimotor and supplementary motor activity and a cere-bellar activation plus some thalamic frontal and cingulatedclusters The mean time span shows that the activity in this


area is less correlated with the stimulation paradigm andwithmotion parameters

The main finding was an increment of activation inmotor areas Indeed the right medial frontal gyrus hasbeen linked to memory retrieval and executive functions Inparticular it has been supposed tomediate attention betweenexternal stimuli and the internally maintained intentionthat is between stimulus-oriented and stimulus-independentprocessing as discussed in [38]

This training required the subject to alternate attentionfrom foot perception and position to the inputs provided byPIGRO and vice versa As far as the premotor areas areconcerned according to Fried et al (1991) as discussed in[39] supplementary and presupplementary motor areas arelinked with the intention and anticipation of the action asdiscussed in [40]

Overall fMRI images are clear and unaffected by thepresence of the prototype in the resonance chamber Theresults also allow understanding the suitability of the methodhere used with the presented device

5 Conclusions

This research studies a locomotor and cognitive trainingusing two mechatronics prototypes In particular the authorsinvestigate the circuits involved in motor imagery and motorlearning at the level of brain plasticity in both healthy subjectsand brain-damaged patients in the future

The experiment was conducted on healthy subjects toassess the possibility of brain reorganization after locomotortraining Sensorimotor training is provided thanks to roboticprototypes developed at the Department of Mechanical andAerospace Engineering Politecnico di Torino

Cognitive training consists of a set of motor imagerytasks Changes in cortical organization are assessed usingfunctionalmagnetic resonance imaging (fMRI) which allowsthe mapping of active processes within the brain thusrevealing the cerebral areas involved in a particular task

In the future various clinical tests on patients shocked byictus and brain event will be carried on using PIGRO

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The work presented in this paper was financed with fundingfrom the Compagnia di San Paolo project ldquoActive exoskele-ton for functional gait rehabilitation of paretic patientsrdquo andwith funding from the Piedmont regional administrationproject entitled ldquoValidation of amethod for gait rehabilitationfor paretic patients using an active orthosisrdquo (2006ndash2008)The authors thank Eng Stefano Cagliero and Eng AnnalisaRigazzi for their help in this research

References

[1] F C Wang C H Yu and T Y Chou ldquoDesign and implemen-tation of robust controllers for a gait trainerrdquo Proceedings of theInstitution of Mechanical Engineers H Journal of Engineering inMedicine vol 223 no 6 pp 687ndash696 2009

[2] D P Ferris G S Sawicki and M A Daley ldquoA physiologistrsquosperspective on robotic exoskeletons for human locomotionrdquoInternational Journal of Humanoid Robotics vol 4 no 3 pp507ndash528 2007

[3] R Gassert E Burdet and K Chinzei ldquoOpportunities andchallenges in MR-compatible roboticsrdquo IEEE Engineering inMedicine and Biology Magazine vol 27 no 3 pp 15ndash22 2008

[4] N V Tsekos A Khanicheh E Christoforou and C MavroidisldquoMagnetic resonance - Compatible robotic and mechatronicssystems for image-guided interventions and rehabilitation areview studyrdquo Annual Review of Biomedical Engineering vol 9pp 351ndash387 2007

[5] R Moser R Gassert E Burdet et al ldquoAnMR compatible robottechnologyrdquo in Proceedings of the IEEE International Conferenceon Robotics and Automation pp 670ndash675 2003

[6] E Burdet R Gassert G Gowrishankar D Chapuis andH Bleuler ldquofMRI compatible haptic interfaces to investigatehumanmotor controlrdquoExperimental Robotics IX vol 21 pp 25ndash34 2006

[7] G Belforte G Eula S Sirolli and S Appendino ldquoDesign andtesting of two mechatronics systems for robotized neuroreha-bilitationrdquo in Proceedings of the 10th International Conferenceon Mechatronics and Precision Engineering Bucarest RomaniaMay 2011

[8] K Sacco S Appendino E Geda et al ldquoDesigning a locomotorand cognitive training with robotic devicesrdquo in Proceedings ofthe EFRR 11th Congress of European Federation for Research inRehabilitation Riva del Garda Italy May 2011

[9] G Belforte G Eula G Quaglia S Appendino F Cauda andK Sacco ldquoMR compatible device for active and passive footmovementsrdquo in Proceedings of the 18th International Workshopon Robotics in Alpe-Adria-Danube Region (RAAD rsquo09) BrasovRomania May 2009

[10] G Belforte and G Eula ldquoOptimisation of a MR-Compatiblemechatronic device useful for fMRI analysisrdquo in Proceedingsof the 21st International Workshop on Robotics in Alpe-Adria-Danube Region (RAAD rsquo12) pp 10ndash13 Naples Italy September2012

[11] G Belforte and G Eula ldquoDesign of an active-passive device forhuman ankle movement during fMRI analysisrdquo Proceedings ofthe Institution ofMechanical Engineers H Journal of Engineeringin Medicine January vol 226 2011

[12] G Belforte G Eula S Appendino and S Sirolli ldquoPneumaticinteractive gait rehabilitation orthosis design and preliminarytestingrdquoProceedings of the Institution ofMechanical EngineersHJournal of Engineering in Medicine vol 225 no 2 pp 158ndash1692011

[13] K Chinzei R Kikinis and F A Jolesz ldquoMR compatibilityof mechatronic devices design criteriardquo in Medical ImageComputing and Computer Assisted Intervention (MICCAI rsquo99)vol 1679 of Lecture Notes in Computer Science pp 1020ndash1031Springer Berlin Germany 1999

[14] R Gassert A Yamamoto D Chapuis L Dovat H Bleulerand E Burdet ldquoActuation methods for applications in MRenvironmentsrdquo Concepts in Magnetic Resonance B MagneticResonance Engineering vol 29 no 4 pp 191ndash209 2006


[15] H Elhawary Z T H Tse A Hamed M Rea B L Daviesand M U Lamperth ldquoThe case for MR-compatible robotics areview of the state of the artrdquo International Journal of MedicalRobotics and Computer Assisted Surgery vol 4 no 2 pp 105ndash113 2008

[16] N Yu C Hollnagel A Blickenstorfer S S Kollias and RRiener ldquoComparison of MRI-compatible mechatronic systemswith hydrodynamic and pneumatic actuationrdquo IEEEASMETransactions on Mechatronics vol 13 no 3 pp 268ndash277 2008

[17] H Elhawary A Zivanovic M Rea et al ldquoA modular approachtoMRI-compatible roboticsrdquo IEEE Engineering inMedicine andBiology Magazine vol 27 no 3 pp 35ndash41 2008

[18] G S Fischer A Krieger I Iordachita C Csoma L L Whit-comb and G Fichtinger ldquoMRI compatibility of robot actuationtechniquesmdasha comparative studyrdquo inMedical Image Computingand Computer-Assisted Intervention (MICCAI rsquo08) vol 5242 ofLecture Notes in Computer Science no 2 pp 509ndash517 SpringerBerlin Germany 2008

[19] C Wienbruch V Candia J Svensson R Kleiser and S SKollias ldquoA portable and low-cost fMRI compatible pneumaticsystem for the investigation of the somatosensensory system inclinical and research environmentsrdquo Neuroscience Letters vol398 no 3 pp 183ndash188 2006

[20] N Yu W Murr A Blickenstorfer S Kollias and R Riener ldquoAnfMRI compatible haptic interface with pneumatic actuationrdquoin Proceedings of the IEEE 10th International Conference onRehabilitation Robotics (ICORR rsquo07) pp 714ndash720 NoordwijkThe Netherlands June 2007

[21] C Raoufi A A Goldenberg and W Kucharczyk ldquoA newhydraulicallypneumatically actuated mrcompatible robot forMRI-guided neurosurgeryrdquo in Proceedings of the 2nd Interna-tional Conference on Bioinformatics and Biomedical Engineering(ICBBE rsquo08) pp 2232ndash2235 Shanghai China May 2006

[22] B J MacIntosh R Mraz N Baker F Tam W R Staines andS J Graham ldquoOptimizing the experimental design for ankledorsiflexion fMRIrdquo NeuroImage vol 22 no 4 pp 1619ndash16272004

[23] S Francis X Lin S Aboushoushah et al ldquofMRI analysis ofactive passive and electrically stimulated ankle dorsiflexionrdquoNeuroImage vol 44 no 2 pp 469ndash479 2009

[24] ISO 7250-1 Basic human bodymeasurements for technologicaldesign Part 1 Body measurement definitions and landmarks

[25] ISOTR 7250-2 Basic human body measurements for techno-logical design Part 2 Statistical summaries of body measure-ments from individual ISO populations

[26] K Kubo T Miyoshi A Kanai and K Terashima ldquoGait reha-bilitation device in central nervous system disease a reviewrdquoJournal of Robotics vol 2011 Article ID 348207 14 pages 2011

[27] I Dıaz G G Gil and E Sanchez ldquoLower-limb robotic rehabili-tation literature review and challengesrdquo Journal of Robotics vol2011 Article ID 759764 11 pages 2011

[28] G S Sawicki K E Gordon and D P Ferris ldquoPoweredlower limb orthoses Applications in motor adaptation andrehabilitationrdquo in 2Proceedings of the IEEE 9th InternationalConference on Rehabilitation Robotics (ICORR rsquo05) pp 206ndash211Chicago Ill USA July 2005

[29] P Beyl M van Damme R van Ham and D Lefeber ldquoDesignand control concepts of an exoskeleton for gait rehabilitationrdquo inProceedings of the 2nd Biennial IEEERAS-EMBS InternationalConference on Biomedical Robotics andBiomechatronics (BioRobrsquo08) pp 103ndash108 Scottsdale Ariz USA October 2008

[30] D Surdilovic J Zhang and R Bernhardt ldquoSTRING-MANwire-robot technology for safe flexible and human-friendly gaitrehabilitationrdquo in Proceedings of the IEEE 10th InternationalConference on Rehabilitation Robotics (ICORR rsquo07) pp 446ndash453 Noordwijk The Netherlands June 2007

[31] X Zhang C Yang J Zhang and Y Chen ldquoA novel DGObased on pneumatic exoskeleton leg for locomotor trainingof paraplegic patientsrdquo in Intelligent Robotics and ApplicationsLecture Notes in Computer Science pp 528ndash537 SpringerBerlin Germany 2008

[32] G Belforte G Eula S Appendino G C Geminiani and MZettin ldquoTutore attivo per neuroriabilitazione motoria degli artiinferiori sistema comprendente tale tutore e procedimento peril funzionamento di tale sistemardquo Patent TO2012A000226 2012

[33] J Perry Gait AnalysismdashNormal and Pathological FunctionSLACK Incorporated 1992

[34] S Ionta A Ferretti A Merla A Tartaro and G L RomanildquoStep-by-step the effects of physical practice on the neuralcorrelates of locomotion imagery revealed by fMRIrdquo HumanBrain Mapping vol 31 no 5 pp 694ndash702 2010

[35] F Malouin and C L Richards ldquoMental practice for relearninglocomotor skillsrdquo Physical Therapy vol 90 no 2 pp 240ndash2512010

[36] J Talairach and P Tournoux Co-Planar Stereotaxic Atlas of theHumanBrain 3-Dimensional Proportional System AnApproachto Cerebral Imaging Thieme Stuttgart Germany 1988

[37] G M Boynton S A Engel G H Glover and D J HeegerldquoLinear systems analysis of functional magnetic resonanceimaging in human V1rdquo Journal of Neuroscience vol 16 no 13pp 4207ndash4221 1996

[38] O Baumann and M W Greenlee ldquoEffects of attention toauditory motion on cortical activations during smooth pursuiteye trackingrdquo PLoS ONE vol 4 no 9 Article ID e7110 2009

[39] I Fried A Katz G McCarthy et al ldquoFunctional organizationof human supplementary motor cortex studies by electricalstimulationrdquo Journal of Neuroscience vol 11 no 11 pp 3656ndash3666 1991

[40] K Sacco F Cauda S Duca et al ldquoA combined robotic andcognitive training for locomotor rehabilitation evidences ofcerebral functional reorganization in two chronic traumaticbrain injured patientsrdquo Frontiers in Human Neuroscience pp 1ndash9 2011

Submit your manuscripts athttpwwwhindawicom

VLSI Design

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of

RotatingMachinery


Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014


Shock and Vibration


Mechanical Engineering

Advances in


Civil EngineeringAdvances in

Acoustics and VibrationAdvances in



Electrical and Computer Engineering

Journal of


Distributed Sensor Networks


The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of


Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014


Active and Passive Electronic Components


Chemical EngineeringInternational Journal of

Control Scienceand Engineering

Journal of


Antennas andPropagation




Navigation and Observation


Advances inOptoElectronics


Volume 2014

RoboticsJournal of



is an interesting research on human motor cortex function-ing

2 Robotic Machines Prototypes

21 BraDiPO BraDiPO [7 9ndash11] is an MR-compatibledevice with 1 DoF (degree of freedom) in the sagittal planewhich allows the patientrsquos ankle joint rotation around itsproper axis It has two pedals to which the patientrsquos feetare secured moved by a pneumatic actuator (Figures 1(a)and 1(b)) and it operates either in ldquoactive moderdquo (patientfollows the pedal movement passively) or in ldquopassive moderdquo(patient moves the feet autonomously and an optical angularpotentiometer put on the pedals records the traveled angle)

Some similar examples are shown in literature as dis-cussed in [13ndash23]

These devices are always MR compatible as they haveto work in the magnetic resonance chamber without pro-ducing interference [13ndash15] and use in general pneumatic orhydraulic controls

BraDiPO can be adapted to patientrsquos feet with ananthropometric range between 10ile woman and 95ileman in agreement with the standards underlined in [24 25]All electrical parts and the PC are remotely located A user-friendly graphical interface is also provided

A flow chart of the BraDiPO management software isshown in Figure 1(c) It was developed by the authors inorder to optimize this investigation Each test starts with ageneral system check after which the operator can choosetest parameters (cylinder velocity pedal angle range andoperating frequency) Each frame has a virtual emergencyswitch to stop software during analysis

Themain problems encountered with the first BraDiPOprototype (Figure 1(b)) are referred to the pneumatic actua-tor position improvement the device position adjustment onthe scanning table the dynamics signal transmission outside-inside from the magnetic resonance chamber

Consequently a second prototype was designed as alsodiscussed in [10] In this second solution the pneumaticactuator (1) is located inside the device (Figure 2(a)) and itis connected to a bar (2) which rotates around hinge O (3) tomove the pedals (4)

As it was found during the initial fMRI analysis withBraDiPO the patientrsquos foot movement could interfere withthe test by producing involuntary head movements

So the device was raised as illustrated in Figure 2(b)because this configuration reduces the transmission of move-ment from patientrsquos feet to head and solves the problem

As the raising of the patientrsquos legs also requires a horizon-tal adjustment of the patientrsquos position on the scanning table(Figure 2(c)) another pneumatic actuator was added underthe box and used asmotor of a horizontal positioning system

The final second BraDiPO prototype is shown in Fig-ure 3

Moreover the problem of a proper dynamics signaltransmission through the long tubes here used (about 10meter) widely presented in literature [16ndash23] for other similardevices and by authors in [11] was studied and already solvedin the first BraDiPO prototype

Here the problem was addressed by selecting nonmag-netic commercial valves and placing them close to thecylinder (PF and PD in Figure 4(a))

In particular this study was carried out using commercialsoftware where the layout of the circuit shown in Figure 4(b)was designed

In this way authors investigated the dynamics perfor-mances of BraDiPO control circuit in order to obtain forthe patientrsquos feet movement the required working frequency(about 1Hz) It is imposed by paradigms connected to thehuman motor cortex activation

In this numerical model a number of parameters werechanged in order to analyze the systemrsquos dynamic behaviorvarying pedal loads (m) working frequencies (f ) supplypressures (119901

119904) and duty cycles (d) The latter parameter

is very important because it directly affects the line sup-plydischarge time

During simulation parameters were varied in the fol-lowing ranges mass m = 1-2 kg working frequency f = 02ndash07Hz duty cycle d = 015 025 040 050 and relative supplypressure 119901

119904= 30lowast105 35lowast105 40lowast105 Pa

In order to validate the numerical model elaborated anexperimental test bench was constructed and some tests werecarried out to compare experimental and theoretical results[11]

An example of results obtained from this study is shownin Figure 5 with 2 kg load 35lowast105 Pa supply pressure 05Hzfrequency and 025 duty cycle [11]

These graphs illustrate a good correspondence betweentheoretical and experimental results carried out in the sametest condition In fact varying the various parameters theshape and the amplitude of the signals are always good andshow a proper functioning of the numericalmodel elaboratedby the authors as can be widely checked in [11]

In particular analysing Figure 5 Pc (see also Figure 4(a))reaches the same maximum amplitude value both in thetheoretical and experimental tests Puf (Figure 4(a)) shows atheoretical amplitude value lower than the experimental onePud (Figure 4(a)) shows the same behaviour in both cases

This fact is certainly due to the little errors that sometimesoccur between numerical models and experimental testsThe main causes here considered are cylinder and valvesfrictionmodeling and physical modeling of a long pneumaticline Overall results obtained are good and give possibilityto construct a final prototype and circuit working alwaysproperly

22 PIGRO The other prototype here involved is calledPIGRO and it is an active exoskeleton for lower limbs withan electropneumatic control

The literature on active lower-limb exoskeletons used forgait rehabilitation as discussed in [26ndash31] chiefly deals withmachines tethered to a fixed stationThey are generally ratherrigid heavy devices and are mostly employed for treadmilltrainingThese systems almost never provide ankle actuation

By contrast PIGRO system is an innovative devicedesigned for the rehabilitation of patients with clinical prob-lems such as hemiplegia tetraparesis and hemiparesis aswell


(a)

8

16

3

5

2

7

4

(b)

Start

Autodiagnosis

No

Yes


Yes

No

Tests

Frequency pause


Save fileSave file

Emergency

Emergency

Yes

No

No

Yes


(c)



1 2

3 4

A

Box

O

H

120576

(a)

z

yh

120572

(b)

120572120572998400

Δy

Δz

(c)









10ilewoman (mm)

95ile man(mm)









Patient


SE

PE

E A F D

P

OE

R1

R2RE




PD

PF

F

Rf1

Rf2

Pc

Puf

T1

T3 T4

T2

Pud

24V

(a)

Rf1

E

F DVc

A

Rf2

PcPF

PuF PuD

PD

Pilot stage PF

Cylinder

Gasdata

Pilot stage PD

CpCp

PPP

CpCpCpCp

A

P

P

T

A

P T

A

P T

A

P

P

T

T AP

TAP

T

M F

k

k k

kkElectric input

+rarr

M+rarrM

X X

+rarr

(b)



000510152025303540

000 060 120 180 240 300 360Time (s)

Pc (b

ar)

0

5

10

15

20

25

30

Elec

tric

inpu

t (V

)


00

02

04

06

08

10

12

000 064 128 192 256 320 384Time (s)

Puf

(bar

)

00

02

04

06

08

10

12

Pud

(bar

)



000 060 120 180 240 300 360Time (s)

m = 2kg f = 05Hz d = 05 ps = 35bar

m = 2kg ps = 35bar f = 05Hz d = 05

m = 2kg ps = 35bar f = 05Hz d = 05

(a)

000510152025303540

000 048 096 144 192 240 288 336 384Time (s)

Pc (b

ar)

000 052 104 156 208 260 312 364Time (s)

000 052 104 156 208 260 312 364Time (s)

00

02

04

06

08

10

12

Puf (

bar)

00

02

04

06

08

10

12

Pud

(bar

)

m = 2kg f = 05Hz d = 05 ps = 35bar

m = 2kg f = 05Hz d = 05 ps = 35bar

m = 2kg f = 05Hz d = 05 ps = 35bar

(b)











(a)

(b) (c)

(d) (e)








05

10152025

Time (s)

60 66 73 79 86 92 98 105 111 118

Left

hip

angl

e (de

g)

minus5

minus10

minus15

minus20

minus25

InputOutput

(a)

0

10

20

30

40

50

60

70

60 66 73 79 86 92 98 105 111 118

Left

knee

angl

e (de

g)

Time (s)

InputOutput

(b)

Time (s)InputOutput

60 66 73 79 86 92 98 105 111 118

Left

ankl

e ang

le (d

eg)

05

101520

minus5

minus10

minus15

minus20

minus25

minus30

(c)






























119883

coor119884

coor119885



















5 Conclusions







Acknowledgments


References











































VLSI Design



RotatingMachinery





Shock and Vibration



Advances in







Journal of





SensorsJournal of








Journal of










Volume 2014

RoboticsJournal of



(a)

8

16

3

5

2

7

4

(b)

Start

Autodiagnosis

No

Yes


Yes

No

Tests

Frequency pause


Save fileSave file

Emergency

Emergency

Yes

No

No

Yes


(c)



1 2

3 4

A

Box

O

H

120576

(a)

z

yh

120572

(b)

120572120572998400

Δy

Δz

(c)









10ilewoman (mm)

95ile man(mm)









Patient


SE

PE

E A F D

P

OE

R1

R2RE




PD

PF

F

Rf1

Rf2

Pc

Puf

T1

T3 T4

T2

Pud

24V

(a)

Rf1

E

F DVc

A

Rf2

PcPF

PuF PuD

PD

Pilot stage PF

Cylinder

Gasdata

Pilot stage PD

CpCp

PPP

CpCpCpCp

A

P

P

T

A

P T

A

P T

A

P

P

T

T AP

TAP

T

M F

k

k k

kkElectric input

+rarr

M+rarrM

X X

+rarr

(b)



000510152025303540

000 060 120 180 240 300 360Time (s)

Pc (b

ar)

0

5

10

15

20

25

30

Elec

tric

inpu

t (V

)


00

02

04

06

08

10

12

000 064 128 192 256 320 384Time (s)

Puf

(bar

)

00

02

04

06

08

10

12

Pud

(bar

)



000 060 120 180 240 300 360Time (s)

m = 2kg f = 05Hz d = 05 ps = 35bar

m = 2kg ps = 35bar f = 05Hz d = 05

m = 2kg ps = 35bar f = 05Hz d = 05

(a)

000510152025303540

000 048 096 144 192 240 288 336 384Time (s)

Pc (b

ar)

000 052 104 156 208 260 312 364Time (s)

000 052 104 156 208 260 312 364Time (s)

00

02

04

06

08

10

12

Puf (

bar)

00

02

04

06

08

10

12

Pud

(bar

)

m = 2kg f = 05Hz d = 05 ps = 35bar

m = 2kg f = 05Hz d = 05 ps = 35bar

m = 2kg f = 05Hz d = 05 ps = 35bar

(b)











(a)

(b) (c)

(d) (e)








05

10152025

Time (s)

60 66 73 79 86 92 98 105 111 118

Left

hip

angl

e (de

g)

minus5

minus10

minus15

minus20

minus25

InputOutput

(a)

0

10

20

30

40

50

60

70

60 66 73 79 86 92 98 105 111 118

Left

knee

angl

e (de

g)

Time (s)

InputOutput

(b)

Time (s)InputOutput

60 66 73 79 86 92 98 105 111 118

Left

ankl

e ang

le (d

eg)

05

101520

minus5

minus10

minus15

minus20

minus25

minus30

(c)






























119883

coor119884

coor119885



















5 Conclusions







Acknowledgments


References











































VLSI Design



RotatingMachinery





Shock and Vibration



Advances in







Journal of





SensorsJournal of








Journal of










Volume 2014

RoboticsJournal of



1 2

3 4

A

Box

O

H

120576

(a)

z

yh

120572

(b)

120572120572998400

Δy

Δz

(c)









10ilewoman (mm)

95ile man(mm)









Patient


SE

PE

E A F D

P

OE

R1

R2RE




PD

PF

F

Rf1

Rf2

Pc

Puf

T1

T3 T4

T2

Pud

24V

(a)

Rf1

E

F DVc

A

Rf2

PcPF

PuF PuD

PD

Pilot stage PF

Cylinder

Gasdata

Pilot stage PD

CpCp

PPP

CpCpCpCp

A

P

P

T

A

P T

A

P T

A

P

P

T

T AP

TAP

T

M F

k

k k

kkElectric input

+rarr

M+rarrM

X X

+rarr

(b)



000510152025303540

000 060 120 180 240 300 360Time (s)

Pc (b

ar)

0

5

10

15

20

25

30

Elec

tric

inpu

t (V

)


00

02

04

06

08

10

12

000 064 128 192 256 320 384Time (s)

Puf

(bar

)

00

02

04

06

08

10

12

Pud

(bar

)



000 060 120 180 240 300 360Time (s)

m = 2kg f = 05Hz d = 05 ps = 35bar

m = 2kg ps = 35bar f = 05Hz d = 05

m = 2kg ps = 35bar f = 05Hz d = 05

(a)

000510152025303540

000 048 096 144 192 240 288 336 384Time (s)

Pc (b

ar)

000 052 104 156 208 260 312 364Time (s)

000 052 104 156 208 260 312 364Time (s)

00

02

04

06

08

10

12

Puf (

bar)

00

02

04

06

08

10

12

Pud

(bar

)

m = 2kg f = 05Hz d = 05 ps = 35bar

m = 2kg f = 05Hz d = 05 ps = 35bar

m = 2kg f = 05Hz d = 05 ps = 35bar

(b)











(a)

(b) (c)

(d) (e)








05

10152025

Time (s)

60 66 73 79 86 92 98 105 111 118

Left

hip

angl

e (de

g)

minus5

minus10

minus15

minus20

minus25

InputOutput

(a)

0

10

20

30

40

50

60

70

60 66 73 79 86 92 98 105 111 118

Left

knee

angl

e (de

g)

Time (s)

InputOutput

(b)

Time (s)InputOutput

60 66 73 79 86 92 98 105 111 118

Left

ankl

e ang

le (d

eg)

05

101520

minus5

minus10

minus15

minus20

minus25

minus30

(c)






























119883

coor119884

coor119885



















5 Conclusions







Acknowledgments


References











































VLSI Design



RotatingMachinery





Shock and Vibration



Advances in







Journal of





SensorsJournal of








Journal of










Volume 2014

RoboticsJournal of



Patient


SE

PE

E A F D

P

OE

R1

R2RE




PD

PF

F

Rf1

Rf2

Pc

Puf

T1

T3 T4

T2

Pud

24V

(a)

Rf1

E

F DVc

A

Rf2

PcPF

PuF PuD

PD

Pilot stage PF

Cylinder

Gasdata

Pilot stage PD

CpCp

PPP

CpCpCpCp

A

P

P

T

A

P T

A

P T

A

P

P

T

T AP

TAP

T

M F

k

k k

kkElectric input

+rarr

M+rarrM

X X

+rarr

(b)



000510152025303540

000 060 120 180 240 300 360Time (s)

Pc (b

ar)

0

5

10

15

20

25

30

Elec

tric

inpu

t (V

)


00

02

04

06

08

10

12

000 064 128 192 256 320 384Time (s)

Puf

(bar

)

00

02

04

06

08

10

12

Pud

(bar

)



000 060 120 180 240 300 360Time (s)

m = 2kg f = 05Hz d = 05 ps = 35bar

m = 2kg ps = 35bar f = 05Hz d = 05

m = 2kg ps = 35bar f = 05Hz d = 05

(a)

000510152025303540

000 048 096 144 192 240 288 336 384Time (s)

Pc (b

ar)

000 052 104 156 208 260 312 364Time (s)

000 052 104 156 208 260 312 364Time (s)

00

02

04

06

08

10

12

Puf (

bar)

00

02

04

06

08

10

12

Pud

(bar

)

m = 2kg f = 05Hz d = 05 ps = 35bar

m = 2kg f = 05Hz d = 05 ps = 35bar

m = 2kg f = 05Hz d = 05 ps = 35bar

(b)











(a)

(b) (c)

(d) (e)








05

10152025

Time (s)

60 66 73 79 86 92 98 105 111 118

Left

hip

angl

e (de

g)

minus5

minus10

minus15

minus20

minus25

InputOutput

(a)

0

10

20

30

40

50

60

70

60 66 73 79 86 92 98 105 111 118

Left

knee

angl

e (de

g)

Time (s)

InputOutput

(b)

Time (s)InputOutput

60 66 73 79 86 92 98 105 111 118

Left

ankl

e ang

le (d

eg)

05

101520

minus5

minus10

minus15

minus20

minus25

minus30

(c)






























119883

coor119884

coor119885



















5 Conclusions







Acknowledgments


References











































VLSI Design



RotatingMachinery





Shock and Vibration



Advances in







Journal of





SensorsJournal of








Journal of










Volume 2014

RoboticsJournal of



000510152025303540

000 060 120 180 240 300 360Time (s)

Pc (b

ar)

0

5

10

15

20

25

30

Elec

tric

inpu

t (V

)


00

02

04

06

08

10

12

000 064 128 192 256 320 384Time (s)

Puf

(bar

)

00

02

04

06

08

10

12

Pud

(bar

)



000 060 120 180 240 300 360Time (s)

m = 2kg f = 05Hz d = 05 ps = 35bar

m = 2kg ps = 35bar f = 05Hz d = 05

m = 2kg ps = 35bar f = 05Hz d = 05

(a)

000510152025303540

000 048 096 144 192 240 288 336 384Time (s)

Pc (b

ar)

000 052 104 156 208 260 312 364Time (s)

000 052 104 156 208 260 312 364Time (s)

00

02

04

06

08

10

12

Puf (

bar)

00

02

04

06

08

10

12

Pud

(bar

)

m = 2kg f = 05Hz d = 05 ps = 35bar

m = 2kg f = 05Hz d = 05 ps = 35bar

m = 2kg f = 05Hz d = 05 ps = 35bar

(b)











(a)

(b) (c)

(d) (e)








05

10152025

Time (s)

60 66 73 79 86 92 98 105 111 118

Left

hip

angl

e (de

g)

minus5

minus10

minus15

minus20

minus25

InputOutput

(a)

0

10

20

30

40

50

60

70

60 66 73 79 86 92 98 105 111 118

Left

knee

angl

e (de

g)

Time (s)

InputOutput

(b)

Time (s)InputOutput

60 66 73 79 86 92 98 105 111 118

Left

ankl

e ang

le (d

eg)

05

101520

minus5

minus10

minus15

minus20

minus25

minus30

(c)






























119883

coor119884

coor119885



















5 Conclusions







Acknowledgments


References











































VLSI Design



RotatingMachinery





Shock and Vibration



Advances in







Journal of





SensorsJournal of








Journal of










Volume 2014

RoboticsJournal of



(a)

(b) (c)

(d) (e)








05

10152025

Time (s)

60 66 73 79 86 92 98 105 111 118

Left

hip

angl

e (de

g)

minus5

minus10

minus15

minus20

minus25

InputOutput

(a)

0

10

20

30

40

50

60

70

60 66 73 79 86 92 98 105 111 118

Left

knee

angl

e (de

g)

Time (s)

InputOutput

(b)

Time (s)InputOutput

60 66 73 79 86 92 98 105 111 118

Left

ankl

e ang

le (d

eg)

05

101520

minus5

minus10

minus15

minus20

minus25

minus30

(c)






























119883

coor119884

coor119885



















5 Conclusions







Acknowledgments


References











































VLSI Design



RotatingMachinery





Shock and Vibration



Advances in







Journal of





SensorsJournal of








Journal of










Volume 2014

RoboticsJournal of



05

10152025

Time (s)

60 66 73 79 86 92 98 105 111 118

Left

hip

angl

e (de

g)

minus5

minus10

minus15

minus20

minus25

InputOutput

(a)

0

10

20

30

40

50

60

70

60 66 73 79 86 92 98 105 111 118

Left

knee

angl

e (de

g)

Time (s)

InputOutput

(b)

Time (s)InputOutput

60 66 73 79 86 92 98 105 111 118

Left

ankl

e ang

le (d

eg)

05

101520

minus5

minus10

minus15

minus20

minus25

minus30

(c)






























119883

coor119884

coor119885



















5 Conclusions







Acknowledgments


References











































VLSI Design



RotatingMachinery





Shock and Vibration



Advances in







Journal of





SensorsJournal of








Journal of










Volume 2014

RoboticsJournal of

















119883

coor119884

coor119885



















5 Conclusions







Acknowledgments


References











































VLSI Design



RotatingMachinery





Shock and Vibration



Advances in







Journal of





SensorsJournal of








Journal of










Volume 2014

RoboticsJournal of







5 Conclusions







Acknowledgments


References











































VLSI Design



RotatingMachinery





Shock and Vibration



Advances in







Journal of





SensorsJournal of








Journal of










Volume 2014

RoboticsJournal of






























VLSI Design



RotatingMachinery





Shock and Vibration



Advances in







Journal of





SensorsJournal of








Journal of










Volume 2014

RoboticsJournal of


bra. di. po and pigro: innovative devices for motor learning

Documents