Mobility, Accessibility and Safety of People with Cerebral Palsy

Ana Marta Carvalho

, Alireza Asvadi

, Carlos Carona

, Ana Lopes

and Urbano Nunes

Institute of Systems and Robotics, University of Coimbra, DEEC - P´olo II, Coimbra, Portugal

Coimbra Cerebral Palsy Association, Cognitive and Behavioral Center for Research and Intervention,

University of Coimbra, Coimbra, Portugal

Keywords:

Cerebral Palsy, Mobility, Accessibility, Safety, Powered Wheelchair, Improvement, Assistive Navigation.

Abstract:

This research characterizes mobility, accessibility and safety of individuals with severe motor impairment such

as users suffering from Cerebral Palsy (CP). Through the analysis of enabling factors, constraints associated

and the search of possible improvements, it is possible to identify the needs in these ﬁelds and subsequently

develop strategies accordingly. The sample was collected in Coimbra Cerebral Palsy Association (APCC)

and it included 16 individuals with CP. To these individuals we gave an evaluation protocol with a form with

clinical and sociodemographic data and a questionnaire. The main limiting factors include building/vehicle

access, difﬁculty in reverse drive and lack of safety. The most valued features of a powered wheelchair are

comfort and structure, easy navigation and wheelchair control and safety. The lack of safety in the outdoors

was a relevant limiting factor. Almost all individuals requested improvements of the powered wheelchair. The

most requested improvements were safety related or related with navigation problems. An assistive navigation

solution based on a shared control algorithm is presented, where a powered wheelchair is equipped with the

Kinect sensor, in order to help the user maneuvering the wheelchair safely.

1 INTRODUCTION

Cerebral palsy (CP) is a complex medical and non-

progressive condition, that is characterized by cogni-

tive and motor disturbances, and it is a consequence

of the damage of speciﬁc brain areas caused before,

during or shortly after birth (Koman et al., 2003).

According to the data obtained from the National

Health Interview Survey from 1988 (Health Statis-

tics and Health & Human Services, 1988), CP ap-

pears as the most disabling clinical situation, involv-

ing the largest number of annual medical contacts and

also the largest number of hospital admissions dur-

ing the year. CP is also the most common disability

in childhood and the trend is to increase its preva-

lence over the last decades (Vargus-Adams, 2003).

Due to accessibility and quality improvement of med-

ical care provided to individuals with CP, the aver-

age life expectancy for this group has increased sig-

niﬁcantly. Therefore, before 1950 few people with

CP survived until adulthood and now is expected that

65% to 90% of children with CP can live past adult-

hood (Zaffuto-Sforza, 2005). However, despite the

increasing prevalence of CP, the medical innovation

and development, observed in the 1970s and 1980s,

contributed to a signiﬁcant increase in average life ex-

pectancy, which boosted the research to understand

how the CP can affect the quality of life (QOL) of

these individuals, including their levels of mobility

and participation (Kennes et al., 2002; Wake et al.,

2003). Research results show that children with CP

have a more impaired QOL in all domains when com-

pared with other able-body children (Vargus-Adams,

2003; Varni et al., 2007), but another study concludes

that the QOL of this group is only lower in the phys-

ical domain and not in the psychological and social

domains (Dickinson et al., 2007). The QOL of adults

with CP is signiﬁcantly affected in all domains as-

sessed by The World Health Organization Quality

of Life (WHOQOL-BREF): Physical, Psychological,

Social Relationships and Environment (Carona et al.,

2010). More speciﬁcally, when compared to other

able-body adults, they reported a lower QOL in the

physical domain (mobility) and in the environment

domain (participation and/or opportunities for recre-

ation and leisure and transportation).

This work aims to research new technologies

that may contribute to the mobility, accessibility and

safety improvement of individuals with CP. We aim

to provide results to support the design and develop-

ment of more suitable solutions to improve mobility,

accessibility and safety. With the overall goal to char-

268

Amaral de Carvalho A., Asvadi A., Carona C., Lopes A. and Nunes U..

Mobility, Accessibility and Safety of People with Cerebral Palsy.

DOI: 10.5220/0005185502680275

In Proceedings of the International Conference on Health Informatics (HEALTHINF-2015), pages 268-275

ISBN: 978-989-758-068-0

 2015 SCITEPRESS (Science and Technology Publications, Lda.)

acterize all three factors referred above, this research

work was organized as follows: (1) Characterization

of subject’s mobility, public transportation use and

Human-Machine Interface (HMI); (2) Relation be-

tween HMIs and powered wheelchair steering perfor-

mance; (3) Identiﬁcation of most valued features and

limiting factors in the use of powered wheelchairs; (4)

Analysis of possible solutions to be implemented in a

powered wheelchair. The lack of safety and the difﬁ-

culty in navigating the powered wheelchair were gen-

erally pointed out as the most limiting factors, which

means there is the need to develop more suitable so-

lutions to improve navigation and safety. A solution

to address both of these problems is to install more

sensors in the powered wheelchair, providing addi-

tional information of the environment and to intro-

duce a new navigation system based on a collabora-

tive controller that shares the information between the

user and the machine. In our case, we decided that

Kinect, a sensor that provides 3D information of the

environment, has certain features that makes it a po-

tential choice:

- It is a compact and lightweight sensor which pro-

vides both RGB and range images;

- It gathers 3D information of the powered

wheelchair’s surrounding from a 3D ﬁeld-of view;

- It is a low cost solution;

- It works at a frequency of 30 Hz;

- Operation range acceptable for indoor environ-

ments: from 0.6 to 3.5m. However, the use of

Kinect to provide environment data to a reactive nav-

igator based on a 2D space representation presents

two difﬁculties: (a) the huge amount of data it pro-

vides and (b) the existence of a blind zone both at

short distance and because of the narrow horizon-

tal ﬁeld-of-view (in comparison to laser radial scan-

ners) (Gonzalez-Jimenez et al., 2013). The current re-

search was based on the work done previously under

the research project Interface10 - Emergent Interfaces

for Improving Accessibility of Persons with Cerebral

Palsy (Carona et al., 2012), (Lopes et al., 2013).

2 APPROACH / METHODOLOGY

2.1 Participants

The sample for this study included 16 individuals and

was collected at APCC, between December 2013 and

March 2014, based on the following inclusion crite-

ria: (1) clinical diagnosis of CP; (2) ability to under-

stand questions and provide answers accordingly; (3)

use of a powered wheelchair; (4) minimum age of 15

years. After obtaining APCC’s formal authorization,

participants were selected by their teams of clinical

follow-up, based on the inclusion criteria listed above.

Before ﬁlling out the questionnaires, all participants

gave informed consent.

2.2 Tools

The evaluation protocol of this study was composed

of the following tools:

1. Clinical and sociodemographic data form:

questionnaire ﬁlled out jointly by the researcher and

the technician responsible for monitoring the subject,

which contains the following information: age, gen-

der, type of CP and associated problems;

2. Gross Motor Function Classiﬁcation System

for Cerebral Palsy - GMFCS-CP (Palisano et al.,

1997): grading scale of the degree of impairment,

structured in ﬁve levels, in which the Level 1 is the

lowest and level 5 the highest. The grading is based

on functional limitations, the need of use of mobil-

ity aids or wheelchairs and also on the quality of

movement. Level 1 and 2 are for manual wheelchair

users. Since in this research we are studying powered

wheelchair users, we are only interested in level 3,

4 and 5. Level 3 is for individuals that need to use

a powered wheelchair in more complicated places,

but for short distances or in easier places to navi-

gate, a manual wheelchair will be enough to ensure

their mobility. Level 4 is for users that can only get

their autonomous mobility with the help of a powered

wheelchair. Level 5 is for users whose mobility is se-

riously compromised. Physical problems limit volun-

tary control of the movements and the control of the

head and trunk.

3. Questionnaire for mobility, accessibility and

safety characterization: questionnaire ﬁlled out by

the subject or with the researcher’s help, with mul-

tiple choice questions, organized into four parts: (1)

characterization of (a) subject’s mobility, (b) use of

public transportation and (c) HMIs; (2) characteriza-

tion of the use of assistive mobility technologies; (3)

most valued features and limiting factors in the use of

the powered wheelchair; (4) possible improvements

of the powered wheelchair.

3 RESULTS

3.1 Sample Characterization

In Table 1 we can see the clinical and sociodemo-

graphic characterization of the sample.

The collected sample (n = 16) had an average age

of 29,80. The most observed type of CP was spastic

Mobility,AccessibilityandSafetyofPeoplewithCerebralPalsy

269

Table 1: Clinical and sociodemographic data.

Age (M) 29,80

Gender (n/%)

Female 3/18,75

Male 13/81,25

Type of Cerebral Palsy (n/%)

Spastic 11/68,75

Dystonic 4/25,00

Ataxic 1/6,25

Additional associated Problems (n/%)

No additional associated problems 6/37,50

Visual problems 5/31,25

Epilepsy 2/12,50

Intellectual problems 2/12,50

Hearing problems 2/12,50

Degree of Impairment

(GMFCS-CP) (n/%)

Level III 2/12,50

Level IV 11/68,75

Level V 3/18,75

(68,75%), followed by dystonic (25,00%) and ataxic

(6,25%). Analyzing other health problems associ-

ated with CP (besides motor impairment), 37,50%

of the cases do not present any additional associ-

ated problems, although 31,25% experience visual

deﬁcits, 12,50% intellectual deﬁcits, 12,50% epilepsy

and 12,50% hearing deﬁcits. According to the in-

clusion criteria associated with the use of an assis-

tive technology to improve mobility, most individuals

(68,75%) are level 4 in the grading scale of the degree

of impairment, 18,75% of the individuals are level 5

and the rest (12,50%) of the subjects are level 3.

3.2 Characterization of Accessibility,

Mobility and Support

Almost all individuals with CP (93,75%) reported be-

ing able to move autonomously with the help of a

powered wheelchair. The number of individuals who

reported not using public transportation (although

they wanted to) is very high (87,50%). This some-

how reﬂects the inadequacy in the access to public

transportation, in which individuals with CP are par-

ticularly vulnerable. Finally, a relatively small per-

centage (25,00%) of people in the group of subjects

need HMIs for computer use. These results can be

veriﬁed in Table 2.

Table 2: Characterization of accessibility, mobility and sup-

port.

Yes (n/%) No (n/%)

Do you have the possi-

bility of moving autono-

mously with the powered

wheelchair? 15/93,75 1/6,25

Do you use public

transportation? 2/12,50 14/87,50

Do you use HMIs for

computer use? 4/25,00 12/75,00

3.3 Experience in Using the HMIs and

Powered Wheelchair vs Quality of

Performance

All subjects participating in this study were consid-

ered experienced users steering a powered wheelchair

since they have several years of experience using it.

Table 3: Level of performance with the HMIs.

HMIs for powered wheelchair

navigation (n/%)

Joystick 13/81,25

Pedal Technology 1/6,25

Head interface with sensors 1/6,25

Chin Technology 1/6,25

Level of performance with HMIs for

powered wheelchair navigation (n/%)

Level 3 3/18,75

Level 4 3/18,75

Level 5 10/62,50

HMIs for computer use (n/%)

Chin Pointer 2/50,00

Pedal Technology 1/25,00

SmartNav - Infrared Technology 1/25,00

Missing=12 (n=4)

Level of performance with HMIs

for computer use (n/%)

Level 5 4/100,00

Missing=12 (n=4)

Level of steering performance

in the powered wheelchair (n/%)

Level 3 3/18,75

Level 4 7/43,75

Level 5 6/37,50

Analyzing Table 3, it is possible to see that all the

users of HMIs for computer use and most individu-

als (62,50%) using the HMIs required for steering the

powered wheelchair, considered themselves in a level

5 of performance. Most of the subjects (81,25%) in

this research use a joystick to help the navigation of

HEALTHINF2015-InternationalConferenceonHealthInformatics

270

the powered wheelchair.

Many of the users (43,75%) were classiﬁed as

level 4 in terms of steering performance of the pow-

ered wheelchair and also, that a signiﬁcant percentage

(37,50%) were classiﬁed as level 5.

3.4 Most Valued Features and Limiting

Factors in the Use of Assistive

Technologies for Mobility

In Table 4 we can see that the most valued features

of a powered wheelchair are its comfort and struc-

ture (68,75%), easy navigation and wheelchair con-

trol (56,25%) and ﬁnally, safety (43,75%). The main

limiting factors in a powered wheelchair are the difﬁ-

culty in reverse drive (37,50%) and building/vehicle

access (31,25%). Also 12,50% of the individuals

complained about the lack of safety.

Table 4: Most valued features and limiting factors in a pow-

ered wheelchair.

Most valued features in a wheelchair

Frequency

(n/%)

Comfort/Positioning 11/68,75

Easy navigation and

wheelchair control 9/56,25

Safety 7/43,75

Dimension 2/12,50

Not speciﬁed 2/12,50

Limiting factors of powered

wheelchair use

Frequency

(n/%)

Difﬁculty in reverse drive 6/37,50

Building/vehicle access 5/31,25

Mechanical aspects 3/18,75

Dimension 3/18,75

Safety 2/12,50

Design 1/6,25

Complicated interfaces 1/6,25

Control and navigation of

the wheelchair 1/6,25

Impractical belt 1/6,25

Not speciﬁed 3/18,75

The architectural barriers were identiﬁed as the

main factor limiting the powered wheelchair use at

home (75,00%) and in the outdoors (87,50%). An-

other relevant limiting factor is again the difﬁculty in

reverse drive at home (18,75%) and lack of safety in

public places (31,25%).

The powered wheelchair inadequacy (75,00%),

the limitations of the vehicles (31,25%) and the ve-

hicle adaptation costs (25,00%) on one hand, and the

lack of adapted transports and complexity of its use

(both referred by all the public transportation users),

Table 5: Limiting factors in the use of assistive technologies

for mobility.

Limiting factors of the powered

wheelchair use at home

Frequency

(n/%)

Architectural barriers 12/75,00

Reverse drive 3/18,75

Strain caused by the

powered wheelchair use 2/12,50

Fatigue 1/6,25

Powered wheelchair inadequacy 1/6,25

No limitation 2/12,50

Limiting factors of the

outdoor access

Architectural barriers 14/87,50

Safety 5/31,25

Strain caused by the

powered wheelchair use 1/6,25

Fatigue 1/6,25

No limitation 1/6,25

Limiting factors in the

private transportation use

Powered wheelchair inadequacy 12/75,00

Vehicle limitation 5/31,25

Vehicle adaptation costs 4/25,00

Difﬁculty placing the powered

wheelchair in the vehicle 3/18,75

No limitation 2/12,50

Limiting factors in the public

transportation use

Complexity of the usage 2/100,00

Shortage of transportation 2/100,00

Missing=14 (n=2)

on the other hand, were the main limiting factors men-

tioned for the use of private and public transportation,

respectively.

The results above can be seen in Table 5. We were

able to get a perception of the mobility and accessi-

bility difﬁculties and also concluded that many of the

complaints were related with the lack of safety and

how relevant this issue is or related with the lack of

an appropriate assistive navigation system to help ma-

neuvering the powered wheelchair.

3.5 Possible Improvements of the

Powered Wheelchair

Since one of the goals of this research is to help im-

prove the QOL of people with CP, the subjects of this

study were asked if their powered wheelchair could

be improved in any way, and almost all of them an-

swered positively (93,75%), as seen in Table 6.

In Table 7 is possible to see the most requested im-

provements: aid for reverse drive (40,00%), collision

Mobility,AccessibilityandSafetyofPeoplewithCerebralPalsy

271

Table 6: Powered wheelchair improvement.

Yes (n/%) No (n/%)

The powered wheel-

chair can be impro-

ved in any way? 15/93,75 1/6,25

avoidance (26,67%), a warning during reverse driv-

ing (26,67%), reverse driving information (26,67%),

comfort/structure improvement (20,00%), assistance

in navigation in more complicated places (13,33%).

Table 7: Most requested powered wheelchair improve-

ments.

Possible improvements

Frequency

(n/%)

Aid for reverse drive 6/40,00

Collision avoidance 4/26,67

Warning during reverse

driving

4/26,67

Reverse driving

information

4/26,67

Comfort/structure

improvement

3/20,00

Assistance in navigation

in more complicated places

2/13,33

Missing=1 (n=15)

Table 8 shows the least requested powered

wheelchair improvements. With the results men-

tioned in Table 7, we can conclude that most of the

suggested improvements (57,14%) affect safety and

could be solved by installing more sensors in the pow-

ered wheelchair and by introducing a new navigation

system.

Table 8: Least requested powered wheelchair improve-

ments.

Possible improvements

Frequency

(n/%)

Wheelchair that lifts 1/6,67

Rear camera/mirror 1/6,67

Wheelchair that lies down 1/6,67

Retracting pedal 1/6,67

Bumper (soften ball

impact in football)

1/6,67

Chance of driving the

wheelchair vertically

1/6,67

Flashers 1/6,67

Autonomous wheelchair 1/6,67

Wheelchair with lights 1/6,67

Buttons design improvement 1/6,67

Improvement of speed

control with the joystick

1/6,67

Missing=1 (n=15)

Collaborative Control

Module

Joystick

Reactive Collision

Avoidance

Risk

Assessment

Obstacle

Detection

Multi-level

2.5D data

processing

Mobile Assistive

Robot

Local Perception

Module

Risk

level

Three 2D

scans

Point cloud

v, w

Safe direction

Obstacles z coordinates

v, w

Number of obstacles

Figure 1: Block diagram of the ANS architecture.

Figure 2: X, Y and Z coordinates of the installed Kinect

sensor.

4 ASSISTIVE NAVIGATION

SYSTEM

In this section we describe the proposed solution

for the control and safe navigation of the powered

wheelchair structured in the Assistive Navigation

System (ANS) shown in Figure 1, which integrates

a human-machine collaborative controller. The ANS

requires an effective model of the local environment,

obtained through the use of a Kinect sensor.

4.1 Local Perception

The Local Perception Module is composed by the

multi-level 2.5D data processing and obstacle detec-

tion submodules.

4.1.1 Multi-level 2.5D Data Processing

The 3D depth data (point cloud) receivedfrom Kinect,

is divided into three 2.5D horizontal scans (one for the

top, one for the middle and one for the bottom) to de-

crease the computational complexity. The 2.5D scans

are composed by the minimum measured distance in

HEALTHINF2015-InternationalConferenceonHealthInformatics

272

each column of depth data (Z-array). The minimum

distance for each 2.5D scan is computed by

′

= min(Z

0, j

, Z

1, j

, ..., Z

479, j

) (1)

where j is the respective column number in the depth

image (Rockey, 2013). Y and X locations for the cor-

responding Z elements are respectively provided in

480x1 and 640x1 arrays. Since the robot can only

move in the X-Z plane, the Y (height) coordinate is

ignored (see Figure 2). The data in Z and X arrays in-

dicate the nearest obstacle locations regardless of the

vertical position of obstacles.

An example of the Kinect point cloud and result-

ing 2.5D scans (2.5D scan resulting from the entire

3D ﬁeld-of-view and multilevel 2.5D scans) obtained

by the proposed method, as well as the 2D scan of the

environment provided by a 2D laser range ﬁnder are

shown in Figure 3.

Kinect has the disadvantage of close range blind

spot, and, due to that, it is not reliable when obstacles

are closer than 0.6 meters and it is blind in distances

less than 0.5 meters. The 3D depth data from Kinect is

condensed into three 2.5D scans, corresponding to the

top, middle and bottom volumes. With this methodol-

ogy it is possible to identify obstacles at three differ-

ent height levels, which allows the Mobile Assistive

Robot (MAR) to identify and approach certain ob-

stacles such as tables or desks and at the same time,

avoiding obstacles that could threaten its safety.

4.1.2 Obstacle Detection

The obstacle detection methodology is based on the

VFH method (Borenstein and Koren, 1991). Each of

the three 2.5D scans provided in the previous step is

divided into 5 angular sectors, which are analyzed to

ﬁnd the closest obstacle in each sector. The number

of obstacles and the distance to each obstacle are ob-

tained through the analysis of the 2.5D scans. There-

fore, if an obstacle in a certain sector has a value (dis-

tance) less than a given threshold, it will be consid-

ered as an actual obstacle that can endanger the user

safety, or that can be approached, depending on the

user’s intent and on the 2.5D scan under analysis.

A second analysis is performed to detect the sec-

tor(s) with highest obstacle density. This information

is then provided to the reactive collision avoidance

module, indicating the obstacle weight values of each

sector.

4.2 Collaborative Control

The Collaborative Control Module has a central role

in the ANS. It decides, according to the perceived sit-

uation, whether to give all the power to the user or to

the machine, or to merge user and machine inputs. It

is composed by Risk Assessment and Reactive Colli-

sion Avoidance submodules.

4.2.1 Risk Assessment

This submodule evaluates the current situation and

makes an appropriate decision according to the infor-

mation obtained by the Local Perception module (see

Figure 1). The algorithm selects the sector with the

closest obstacle (the most dangerous sector) among

all sectors, and classiﬁes the current situation accord-

ing to the relative position of the MAR to that sector:

- Obstacles at a distance greater than 0.9 m present

no risk to the user (risk level 0).

- Obstacles that are located between 0.6 and 0.9 m

are classiﬁed as potential obstacles. Obstacles in this

class are later sub-divided in medium risk obstacles,

those located in a distance between 0.75 and 0.9 me-

ters (risk level 1 or 2), and high risk obstacles those lo-

cated in a distance between 0.6 and 0.75 meters (risk

level 3 or 4).

- Obstacles at a distance less than 0.6 m represent

eminent danger (risk level 5).

The current direction taken by MAR also affects

the risk level classiﬁcation. User can move towards

the obstacle or go away from obstacles. The closer

the obstacles are, the greater is the risk.

4.2.2 Reactive Collision Avoidance

The main objective of the Reactive Collision Avoid-

ance submodule is to avoid the obstacles in the vicin-

ity of the MAR. The algorithm takes into account the

number of obstacles in the sectors, the safest direction

to follow, the distance to the obstacle and the angular

and linear speed commands that the user is providing

to the MAR. It acts as a Traded Controller and effec-

tively denies or allows the user commands or acts as

a Shared Controller by combining the robot naviga-

tion commands with the user commands. It assists

user for maneuvering in more complicated situations

as well as avoiding collision in order to achieve a bet-

ter level of safety. The rules composing the reactive

navigation are:

- Traded Control - in this case the steering con-

trol is fully delegated to the user or to the MAR: 1)

When there are no obstacles in risky area, the user

commands are followed, constrained by a maximum

speed value due to safety reasons; 2) If an obstacle is

endangering the user safety, the MAR stops and turns

until a safe direction is found.

- Shared Control - in this case the control is shared

between the user and the MAR: 1) When the user is in

a medium risk situation the maximum linear speed is

Mobility,AccessibilityandSafetyofPeoplewithCerebralPalsy

273

Figure 3: (a) Kinect RGB image, (b) Kinect’s pointcloud and the 2.5D scan resulting from the entire 3D ﬁeld-of-view, (c) the

Kinect’s 2.5D scan resulting from the 3D ﬁeld-of-view, (d) 2D scan from Hokuyo URG-04LX laser scanner with the Kinect’s

ﬁeld-of-view highlighted in yellow, (e) sum of the three level 2.5D scans provided by Kinect, f) the top 2.5D scan, g) the

middle 2.5D scan and h) the bottom 2.5D scan.

reduced for safety reasons. When the user is moving

towards an obstacle the linear speed is even more re-

duced and varies with distance; 2) In a high risk situa-

tion, the reactive navigation will affect both linear and

angular speeds. Linear speed is reduced proportion-

ally to the MAR distance to the obstacle. The MAR

rotates towards a safe direction, which is chosen ac-

cording to the sectors with lower obstacle density.

The reactive navigation module was developed in

a way to lead the MAR avoid deadlocks (U-shaped

obstacles), by using a short term memory, which

stores the number of perceivedobstacles in the robot’s

vicinity. RobChair (Lopes et al., 2012), (Lopes et al.,

2013) was the MAR used in the experiments (see Fig-

ure 4). An example of a trajectory performed in a real

test experiment is shown in Figure 5. Some situations

can be underlined:

Situation 1: In this situation, the user successfully

approached the table. The MAR identiﬁed it as a safe

situation since this obstacle had the features (height)

of a table. The control was delegated to the user.

Situation 2, 3 and 4: The user steered the

wheelchair to approach obstacles and the MAR suc-

cessfully avoided them. Because the user was quickly

navigating towards the obstacles, the traded control

was activated almost immediately in order to prevent

a collision. The control was delegated to the MAR.

Situation 5: Shared control was activated since the

user tried to approach the side of a table. Once the

user was navigating slower than in the previous situa-

tions, traded control was not activatedsince the shared

controller commands were able to deviate the MAR

and avoid collision with the obstacle.

5 CONCLUSION AND FUTURE

WORK

This paper presents the characterization of mobility,

Figure 4: RobChair - Mobile Assistive Robot.

Figure 5: MAR navigation trajectory in ISR test scenario.

accessibility and safety in a group of individuals with

Cerebral Palsy. The results suggest that user expe-

rience in steering the powered wheelchair is a deci-

sive factor for a good level of performance. Younger

users can have increased troubles driving the pow-

ered wheelchair due to the lack of experience, there-

fore contributing to a more complex navigation and

compromised safety. Combining these results with

HEALTHINF2015-InternationalConferenceonHealthInformatics

274

the suggested improvements by the users, we con-

cluded that the lack of safety and the difﬁculty in nav-

igating the powered wheelchair were the most limit-

ing factors, in general, and also the most suggested

(57,14%) to be improved. The ANS was tested both

in simulated environment conditions and in real con-

ditions. The promising results show that the collab-

orative controller and other modules of the ANS ar-

chitecture compose a structure on which it is worth

continuing to devote research effort.

Future work includes adding short range sensors

for close distances, improving the collaborative con-

trol methodology to allow safe navigation in human

environments, and also solving other remaining issues

pointed out by powered wheelchair users. Getting a

larger user sample could also point us toward future

research directions.

ACKNOWLEDGEMENTS

This work has been supported by the FCT project

”AMS-HMI2012 - RECI/EEI-AUT/0181/2012” and

project ”ProjB-Diagnosis and Assisted Mobility

- Centro-07-ST24-FEDER-002028” with FEDER

funding, programs QREN and COMPETE.

REFERENCES

Borenstein, J. and Koren, Y. (1991). The vector ﬁeld

histogram-fast obstacle avoidance for mobile robots.

Robotics and Automation, IEEE Transactions on,

7(3):278–288.

Carona, C., Canavarro, M. C., Pereira, M., Vaz-Serra, A.,

Quartilho, M., Paredes, T., Rijo, D., Gameiro, S., and

Sim˜oes, M. (2010). Qualidade de vida de indiv´ıduos

adultos com paralisia cerebral e dos seus cuidadores

familiares. In Qualidade de vida e sa´ude – Uma

abordagem na perspectiva da Organizac¸˜ao Mundial

de Sa´ude (in Portuguese). Fundac¸˜ao Calouste Gul-

benkian.

Carona, C., Reis, P., Almeida, L., Pires, G., Lopes, A.,

Almeida, A., Machado, D., Vaz, L., Moita, F., Carid´a,

V., Castela-Lobo, J., Figueira, A., Antunes, F., Bran-

quinho, A., Elias, C., and Nunes, U. (2012). Mobili-

dade, acessibilidade e participac¸˜ao em indiv´ıduos com

paralisia cerebral (in portuguese). Technical report,

Institute of Systems and Robotics, Coimbra Cerebral

Palsy Association, Cognitive and Behavioural Center

for Research and Intervention - University of Coim-

bra.

Dickinson, H. O., Parkinson, K. N., Ravens-Sieberer, U.,

Schirripa, G., Thyen, U., Arnaud, C., Beckung, E.,

Fauconnier, J., McManus, V., Michelsen, S. I., Parkes,

J., and Colver, A. F. (2007). Self-reported qual-

ity of life of 8-12-year-old children with cerebral

palsy: a cross-sectional european study. Lancet,

369(9580):2171–2178.

Gonzalez-Jimenez, J., Ruiz-Sarmiento, J., and Galindo, C.

(2013). Improving 2d reactive navigators with kinect.

In Int. Conf. on Informatics in Control, Automation

and Robotics, ICINCO 2013.

Health Statistics, N. C. f. and Health & Human Services, U.

S. D. (1988). National health interview survey: Child

health supplement.

Kennes, J., Rosenbaum, P., Hanna, S. E., Walter, S., Rus-

sell, D., Raina, P., Bartlett, D., and Galuppi, B.

(2002). Health status of school-aged children with

cerebral palsy: information from a population-based

sample. Developmental Medicine & Child Neurology,

44(4):240–247.

Koman, L. A., Smith, B. P., and Balkrishnan, R. (2003).

Spasticity associated with cerebral palsy in children:

guidelines for the use of botulinum a toxin. Pediatric

Drugs, 5(1):11–23.

Lopes, A., Pires, G., and Nunes, U. (2012). Robchair:

Experiments evaluating brain-computer interface to

steer a semi-autonomous wheelchair. In Intelligent

Robots and Systems (IROS), 2012 IEEE/RSJ Interna-

tional Conference on, pages 5135–5136.

Lopes, A. C., Pires, G., and Nunes, U. (2013). Assisted

navigation for a brain-actuated intelligent wheelchair.

Robot. Auton. Syst., 61.

Palisano, R., Rosenbaum, P., Walter, S., Russell, D., Wood,

E., and Galuppi, B. (1997). Gross motor function clas-

siﬁcation system. Developmental Medicine & Child

Neurology, 39(4):214–223.

Rockey, C. (2013). depthimage

to laserscan - ros wiki.

Vargus-Adams, J. (2003). Health-related quality of life

in childhood cerebral palsy. Archives of Physical

Medicine Rehabilitation, 86(5):940–945.

Varni, J. W., Limbers, C. A., and Burwinkle, T. M. (2007).

Impaired health related quality of life in children

and adolescents with chronic conditions: A compar-

ative analysis of 10 disease clusters and 33 disease

categories/severities utilizing the pedsql 4.0 generic

core scales. Health and Quality of Life Outcomes,

5(43):673–681.

Wake, M., Salmon, L., and Reddihough, D. (2003). Health

status of australian children with mild to severe cere-

bral palsy: cross-sectional survey using the child

health questionnaire. Developmental Medicine and

Child Neurology, 45(3):194–199.

Zaffuto-Sforza, C. D. (2005). Aging with cerebral palsy.

Physical Medicine and Rehabilitation Clinics of North

America, 16(1):235–249.

Mobility,AccessibilityandSafetyofPeoplewithCerebralPalsy

275