A TECHNOLOGICAL AND STATISTICAL STATE-OF-THE-ART

STUDY REGARDING ACTIVE MOTION-ORIENTED ASSISTIVE

DEVICES

Daniel Pina

, Ant

onio Augusto Fernandes

1,2

, Joaquim Gabriel Mendes

1,2

and Renato Natal Jorge

1,2

Instituto de Engenharia Mec

anica, Porto, Portugal

Faculdade de Engenharia da Universidade do Porto, Porto, Portugal

Keywords:

Powered exoskeleton, Active orthosis, Rehabilitation robot.

Abstract:

Active orthoses and powered exoskeletons, among other denominations, are devices made to attach to one

or several human limbs in order to assist or replace its wearer’s movement through means of electronically-

controlled actuators and/or mechanical brakes. The technology developed for these devices can be used for

rehabilitation, general strength enhancement for industrial or military purposes, among other situations. In or-

der to create a comprehensive state-of-the-art work for this class of devices, several online scientiﬁc databases

were used to gather articles related to this subject. Afterwards, a custom database was created to contain,

organize and cross the information gathered from each relevant article. This work presents statistical results

regarding the actuation technologies, the man-machine interface sensors and the corresponding interpretation

algorithms. There is also a brief study about the localization of the scientiﬁc research, according to the tar-

geted body part of the active device. The results show that the DC Motor is, by a wide margin, the most used

actuator technology. This margin is reduced when wearable devices with weight constraints are developed.

The electromyographic sensors are the most widely used sensors, but when these are grouped into physical

variable classes, the force-related sensors show a higher number of occurrences. Regarding the processing

algorithms required for the man-machine interface, it is often required to develop a custom algorithm for these

devices.

1 INTRODUCTION

Active orthoses and powered exoskeletons, among

other denominations, are devices made to attach to

one or several human limbs in order to assist its

wearer’s movement through means of electronically-

controlled actuators and/or mechanical brakes. The

technology developed for these devices can be used

for various purposes, such as rehabilitation (Jia-fan

et al., 2010; Yuanjie Fan, 2009), industrial (Ya-

suhisa Hasegawa, 2010; Low et al., 2005) and general

strength enhancement (Kazerooni and Steger, 2006;

Cao et al., 2009a).

As seen in Figure 2, from 2005 onwards, the

global scientiﬁc community has shown a large in-

crease in the overall interest around these devices.

In order to create a comprehensive and organized

state-of-the-art bibliographic work, several articles re-

lated to active motion-oriented assistive devices, re-

gardless of their purpose, were gathered from seven

scientiﬁc databases.

There is a lack of a consistent naming convention

for these devices, creating some difﬁculties in the

task of gathering scientiﬁc bibliography over this sub-

ject. Depending on various authors and associated re-

search institutions, several different names have been

given to the same kind of devices, regardless of their

purpose, technology involved or target body parts.

Therefore, in order to gather as much articles related

to this subject as possible, the search was repeated

for each of the most recurrent terms used to describe

these devices.

Afterwards, a database was created to contain, or-

ganize and cross the information gathered from each

relevant article.

This work presents the statistical results regard-

ing the localization of the scientiﬁc research on these

devices, the actuation technologies, the man-machine

interface sensors and corresponding interpretation al-

gorithms.

245

Pina D., Augusto Fernandes A., Gabriel Mendes J. and Natal Jorge R..

A TECHNOLOGICAL AND STATISTICAL STATE-OF-THE-ART STUDY REGARDING ACTIVE MOTION-ORIENTED ASSISTIVE DEVICES.

DOI: 10.5220/0003795702450250

In Proceedings of the International Conference on Biomedical Electronics and Devices (BIODEVICES-2012), pages 245-250

ISBN: 978-989-8425-91-1

 2012 SCITEPRESS (Science and Technology Publications, Lda.)

2 STATE-OF-THE-ART

DATABASE CONSTRUCTION

In order to gather information into the state-of-the-

art database, the following online search engines from

scientiﬁc research databases were used: ”Science Di-

rect”, ”IEEE Xplore”, ”MetaPress”, ”ACM Digital

Library”, ”ASME Digital Library”, ”IOP Science”

and ”Emerald”.

For each engine, a search was made for each of

the following terms: ”Exoskeleton”, ”Orthosis”, ”Or-

thesis” and ”Rehabilitation Robot”.

From the search results, all the articles containing

a description or a study about an existing active or-

thosis or powered exoskeleton (regardless of its func-

tionality, purpose or prototype stage) were selected.

The articles retrieved were made available under

the ”B-On: University of Porto” program.

This search, conducted during December 2010,

resulted in the retrieval of 203 articles. Out of these,

15 were state-of-the-art and/or comparison articles.

From each of the remaining 188 articles, various tech-

nical and non-technical characteristics were gathered

into a custom-made database, using Microsoft Ofﬁce

Access 2010.

The purpose of this database is to provide statis-

tical studies regarding the scientiﬁc development of

active orthoses and exoskeletons, which can be au-

tomatically updated if more articles are added in the

future. A simpliﬁed representation of the database’s

entity-relationship model can be found in Figure 1:

Figure 1: Simpliﬁed representation of the database’s entity-

relationship model, using Crow’s Foot Notation.

An article is included in the database as part of

a ”Project”, which can represent any device from a

hardware prototype in development stage to a com-

mercial product.

Each project describes the actuator technology

used, the target limbs or body parts, its main and sec-

ondary (if available) purpose, its wearability and the

related articles.

From each article entry, the database also records

its publication, the name of each author and associ-

ated research institution, the ”Input Type” (sensors

used for the user/patient interface, when applicable)

and the ”Input Processing Algorithm” (algorithm that

translates the user/patient’s ”will” to move between

the sensor interface and the computing system, when

applicable).

3 GENERAL STATISTICS

The ﬁrst general statistics that can be taken from the

database is the general commitment from the scien-

tiﬁc community regarding active motion-oriented as-

sistive devices. Figure 2 shows the number of articles

gathered for each corresponding year of ﬁnal draft, as

well the number of related authors.

Figure 2: In blue, the number of articles related to active

orthoses and exoskeletons gathered per year. In red, the

number of authors associated to the gathered articles.

Given that 2005 is the year where a large increase

in scientiﬁc interest was observed, all the statistical

studies presented in the article are, from this point,

calculated using articles from the beginning of 2005

onwards.

A relevant information that can be gathered from

the database is to determine which are the most de-

dicated research institutions for this kind of devices,

depending on the targeted body part. Figure 3 shows

the research institutions with more associated articles

for each targeted body part.

The terms ”Lower Limbs” and ”Upper Limbs” are

used for devices that attach and actuate over more

BIODEVICES 2012 - International Conference on Biomedical Electronics and Devices

246

Figure 3: Number of articles associated to Active Orthoses

or Powered Exoskeletons, depending on the targeted body

part.

than one body part, such as entire limbs. For the

”Hands” and ”Knee”, the graphic considers only re-

search institutions with two or more article associ-

ations. For the other represented body parts, the

graphic considers research institutions with three or

more associations.

Out of the 123 Projects, 89 are primarily dedi-

cated to rehabilitation activities, from retraining mo-

tor movements to reestablishing or aiding motor

movement to target limbs due to injury or stroke.

The remaining 34 Projects are mostly dedicated to

strength and/or endurance enhancement, aimed at var-

ious activities like agriculture or industrial environ-

ments.

4 ACTUATOR TECHNOLOGY

The choice of the Actuator Technology for an active

assistive device depends on various factors, such as

operation noise and safety, energy efﬁciency, control-

lability, among other factors. However, when con-

sidering a portable or ”wearable” device, where the

user must support its weight, the torque/volume and

torque/weight become critical factors. The following

statistics show the proportions of Actuator Technolo-

gies chosen by researchers for active assistive devices

depending on their wearability.

Therefore, the results for the ”Actuator Type” in

each ”Project” were crossed with the ”Wearability”

table (seen in Figure 1) Three types of wearability are

considered:

• Not wearable (Figure 4), where the device is ﬁxed

to a wall or a table, so the weight is less impor-

tant (Worsnopp et al., 2007; Banala et al., 2007).

• Wearable while sitting on a wheelchair (Figure 5),

where the weight of the device is supported by a

wheelchair (Herder, 2005; Rahman et al., 2010).

• Wearable while walking (Figure 6), where the de-

vice allows (or is planned to allow, at some point

in the prototype stage) the user to walk freely

while attached. In this case, the torque-per-weight

ratio holds the most importance since the user

and/or the device itself have to support the in-

creased weight (Cao et al., 2009b; Sankai, 2006).

Figure 4: Proportions of non-wearable active ortho-

sis/exoskeleton projects, grouped by actuator technology.

Figure 5: Proportions of active orthosis/exoskeleton

projects mounted on a wheelchair, grouped by actuator

technology.

The most popular actuator technology used for

creating active orthoses and exoskeletons is the DC

Motor. This actuator technology, and its various im-

plementations, is one of the oldest actuation technolo-

gies used in active devices. Therefore, various and ef-

fective control methods are well studied and pre-built

motor controllers can be achieved from several man-

ufacturers (Motorcontrol, 2011), making this techno-

logy easier to implement from a control standpoint.

However, some newer actuation technologies

(i.e. McKibben actuators) have a substantially larger

torque-per-weight ratio (Plettenburg, 2005), which

A TECHNOLOGICAL AND STATISTICAL STATE-OF-THE-ART STUDY REGARDING ACTIVE

MOTION-ORIENTED ASSISTIVE DEVICES

247

Figure 6: Proportions of wearable active ortho-

sis/exoskeleton projects, grouped by actuator technology.

should provide a practical advantage for weight-

dependent, or ”wearable” orthoses and exoskeletons.

The graphic in Figure 6 shows that the weight

concerns, when projecting a wearable device, are dri-

ving the scientiﬁc community to try other actuation

technologies with more convenient torque-per-weight

ratios, such as McKibben actuators, despite the in-

creased difﬁculty in developing control algorithms for

a stable and safe operation (Daerden, 1999).

5 MAN-MACHINE INTERFACE

5.1 Sensor Technologies

Regarding the following statistics, all the sensor input

types that are used for the man-machine interface be-

tween the user/patient and the device were gathered.

The sensors exclusively used for internal actuator op-

eration generally belong to a different part of the con-

trol loop and are more often documented within the

scientiﬁc studies related to the actuator technology.

As these studies may not relate directly to the de-

velopment of active orthoses and exoskeletons, these

sensors weren’t gathered into the database.

This information, represented as ”Input Type” in

Figure 1, was taken from each article, as different ar-

ticles in the same project may refer to different sensor

input hardware (Carignan et al., 2005; Carignan et al.,

2008).

The sensor technologies were grouped into ma-

jor categories, with each one determining the type of

physical variables each sensor is measuring.

The list of gathered sensor types was divided into

the following classes:

• Electromyographic sensors;

• Force-related sensors: consisted of torque sen-

sors, force sensors, ﬂoor-reaction-force sensors,

force-sensing-resistors and force transducers;

• Angle-related sensors: consisted of potentiome-

ters, rotary encoders, dc motor encoders, go-

niometers and gyroscopes;

• Other sensors: All the sensor technologies with

less than three occurrences like accelerometers,

on/off switches, optical tracking devices, among

others;

• No Input: referring to devices using automated

sequences for rehabilitation/training purposes,

without having a direct connection between the

user/patient and the device (Costa and Caldwell,

2006; Shibata et al., 2010);

• Undisclosed: term applied in the database when

an article doesn’t mention the sensor technol-

ogy used between the user/patient and the device,

which occurs in some articles dedicated to trial

tests and their results (Vanderniepen et al., 2008;

Boehler et al., 2008).

Figure 7: Proportion of sensor technologies, grouped by

physical variable classes.

5.2 Input Processing Algorithms

Another technological feature that can be evaluated

from the database is the ”Input Processing Algo-

rithm”. This is the algorithm used to process the

user’s ”will” to move the actuated limb or body part,

using the digital data retrieved from the sensors dedi-

cated to the man-machine interface.

Figure 8 shows the number of occurrences for the

most popular algorithms for the man-machine inter-

face used in active assistive devices. Like the sensor

technology, different articles belonging to the same

project may refer to experimenting different algo-

rithms while using the same orthosis/exoskeleton pro-

totype (Wege and Hommel, 2006; Wege and Zimmer-

mann, 2007). For this reason, this information was

taken from each article.

BIODEVICES 2012 - International Conference on Biomedical Electronics and Devices

248

Figure 8: Number of occurrences for Input Processing Al-

gorithms, taken from the gathered articles.

The ”Automated Sequence” result refers to the lack

of a man-machine interface, and is therefore related

to the results for ”No Input” in the ”Sensor Input”

data. Algorithms with less than three occurrences

were gathered into a single group called ”Other”.

”Undisclosed Algorithm” is the term applied

when an article doesn’t mention what kind of al-

gorithm the device uses for interpreting the man-

machine interface(Raparelli et al., 2007).

”Custom Algorithm” refers to either a combina-

tion of the other mentioned algorithms or a com-

pletely new formula (Andreasen et al., 2005; Kong

et al., 2009). In these cases, an ”off-the-shelf” solu-

tion isn’t applied.

6 DISCUSSION AND

CONCLUSIONS

As seen in Figure 2, active orthoses and exoskeletons

have suffered a recent increased interest throughout

the scientiﬁc community, starting in 2005. This fact

can be explained through general advances in con-

trol and actuator technologies, the increasingly larger

trend of researching technological solutions for oth-

erwise difﬁcult tasks and the ageing of the population

in certain countries.

The actuator statistics show that the most popular

actuator for active assistive devices is the DC Motor

so far. Although this actuator technology presents a

relatively low torque/weight and torque/volume ratio,

its matured control methods makes it the most widely

adopted choice.

However, when studying a wearable device with

weight and volume constraints, as we see in Fig-

ure 6, the DC Motor is seen to be adopted in less than

half of the studied projects. As actuator technologies

with higher torque/weight and torque/volume ratios

become mature, it may be expected to see DC Motors

being gradually replaced for wearable devices, partic-

ularly by McKibben actuators.

Regarding the statistics relative to the sensors

used, it can be seen that although force-related sensors

are actually being tried out in larger proportions, the

magnitude of occurrences is similar for all the classes.

The statistics relative to the Input Processing Al-

gorithm show that most of the time, a custom-made

algorithm is necessary to develop an efﬁcient compu-

tational method to translate the user/patient’s ”will”

to move the actuated limb or body part. This means

that creating an algorithm to efﬁciently control an ac-

tive orthosis or exoskeleton usually becomes a very

relevant and time-consuming task during the develop-

ment of a functional device.

As for future work, the Authors’ plan is to keep

updating the database on a yearly basis, in order to

follow the technological advances and trends on ac-

tive motion-oriented assistive devices.

REFERENCES

Andreasen, D. S., Allen, S. K., and Backus, D. A. (2005).

Exoskeleton with emg based active assistance for re-

habilitation. In IEEE International Conference on Re-

habilitation Robotics.

Banala, S. K., Kulpe, A., and Agrawal, S. K. (2007). A

powered leg orthosis for gait rehabilitation of motor-

impaired patients (alex). In IEEE International Con-

ference on Robotics and Automation.

Boehler, A. W., Hollander, K. W., Sugar, T. G., and Shin,

D. (2008). Design, implementation and test results

of a robust control method for a powered ankle foot

orthosis (afo). In IEEE International Conference on

Robotics and Automation.

Cao, H., Ling, Z., Zhu, J., Wang, Y., and Wang, W. (2009a).

Design frame of a leg exoskeleton for load-carrying

augmentation. In EEE International Conference on

Robotics and Biomimetics.

Cao, H., Ling, Z., Zhu, J., Wang, Y., and Wang, W. (2009b).

sign frame of a leg exoskeleton for load-carrying

augmentation. In IEEE International Conference on

Robotics and Biomimetics.

Carignan, C., Liszka, M., and Roderick, S. N. (2005). De-

sign of an arm exoskeleton with scapula motion for

shoulder rehabilitation. In International Conference

on Advanced Robotics.

Carignan, C., Naylor, M. P., and Roderick, S. N. (2008).

Controlling shoulder impedance in a rehabilitation

arm exoskeleton. In IEEE International Conference

on Robotics and Automation.

Costa, N. and Caldwell, D. G. (2006). Control of a

biomimetic ”soft-actuated” 10dof lower body ex-

A TECHNOLOGICAL AND STATISTICAL STATE-OF-THE-ART STUDY REGARDING ACTIVE

MOTION-ORIENTED ASSISTIVE DEVICES

249

oskeleton. In International Conference on Biomedical

Robotics and Biomechatronics (BioRob).

Daerden, F. (1999). Conception and Realization of Pleated

Pneumatic Artiﬁcial Muscles and their Use as Com-

pliant Actuation Elements. PhD thesis, Vrije Univer-

siteit Brussel.

Herder, J. L. (2005). Development of a statically balanced

arm support: Armon. In IEEE International Confer-

ence on Rehabilitation Robotics.

Jia-fan, Z., Yi-ming, D., Can-jun, Y., Yu, G., Ying, C.,

and Yin, Y. (2010). 5-link model based gait trajec-

tory adaption control strategies of the gait rehabilita-

tion exoskeleton for post-stroke patients. Mechatron-

ics, 20, Issue 3.

Kazerooni, H. and Steger, R. (2006). The berkeley lower ex-

tremity exoskeleton. Transactions of the ASME, 128.

Kong, K., Moon, H., Hwang, B., Jeon, D., and Tomizuka,

M. (2009). Impedance compensation of subar for

back-drivable force-mode actuation. IEEE Transac-

tions on Robotics, 25:Issue 3.

Low, K. H., Liu, X., and Yu, H. (2005). Development of ntu

wearable exoskeleton system for assistive technolo-

gies. In International Conference on Mechatronics

and Automation.

Motorcontrol (2011). List of compa-

nies providing dc motor drives -

www.motorcontrol.com/2007homepagelinks/dccomp

anies.htm.

Plettenburg, D. H. (2005). Pneumatic actuators, a compar-

ison of energy-to-mass ratios. In Proceedings of the

2005 IEEE 9th International Conference on Rehabili-

tation Robotics.

Rahman, M. H., Saad, M., e, J. P. K., and Archambault, P. S.

(2010). Exoskeleton robot for rehabilitation of elbow

and forearm movements. In Mediterranean Confer-

ence on Control & Automation, Volume 18.

Raparelli, T., Zobel, P. B., Durante, F., Antonelli, M., Rai-

mondi, P., and Costanzo, G. (2007). First clinical in-

vestigation on a pneumatic lumbar unloading orthosis.

In International Conference on Complex Medical En-

gineering IEEE/ICME.

Sankai, Y. (2006). Leading edge of cybernics: Robot suit

hal. In SICE-ICASE International Joint Conference.

Shibata, Y., Imai, S., Nobutomo, T., Miyoshi, T., and ichi-

roh Yamamoto, S. (2010). Development of body

weight support gait training system using antagonis-

tic bi-articular muscle model. In Annual International

Conference of the IEEE EMBS, Volume 32.

Vanderniepen, I., Ham, R. V., Damme, M. V., and Lefeber,

D. (2008). Design of a powered elbow orthosis for

orthopaedic rehabilitation using compliant actuation.

In International Conference on Biomedical Robotics

and Biomechatronics (BioRob).

Wege, A. and Hommel, G. (2006). Development and con-

trol of a hand exoskeleton for rehabilitation of hand

injuries. In Human Interaction with Machines.

Wege, A. and Zimmermann, A. (2007). Electromyography

sensor based control for a hand exoskeleton. In Elec-

tromyography Sensor Based Control for a Hand Ex-

oskeleton.

Worsnopp, T. T., Peshkin, M. . A., Colgate, J. E., and Kam-

per, D. G. (2007). An actuated ﬁnger exoskeleton for

hand rehabilitation following stroke. In IEEE Interna-

tional Conference on Rehabilitation Robotics.

Yasuhisa Hasegawa, Kosuke Watanabe, Y. S. (2010). Per-

formance evaluations of hand and forearm support

system. In Conference on Intelligent Robots and Sys-

tems, IEEE/RSJ International.

Yuanjie Fan, Y. Y. (2009). Mechanism design and mo-

tion control of a parallel ankle joint for rehabilitation

robotic exoskeleton. In IEEE International Confer-

ence on Robotics and Biomimetics.

BIODEVICES 2012 - International Conference on Biomedical Electronics and Devices

250