A

a

a

w

a

a

l

d

m

t

s

n

t

#

K

Q

f

0

d

Navigational strategies during fast walking: A comparison

between trained athletes and non-athletes

Martin Gerin-Lajoie a, Janet L. Ronsky b, Barbara Loitz-Ramage b, Ion Robu b,Carol L. Richards a, Bradford J. McFadyen a,*

a Center for Interdisciplinary Research in Rehabilitation and Social Integration (CIRRIS), Department of Rehabilitation,

Faculty of Medicine, Laval University, Quebec, Canadab Human Performance Laboratory, Faculty of Kinesiology & Department of Mechanical and Manufacturing Engineering,

University of Calgary, Calgary, Alberta, Canada

Received 21 July 2006; received in revised form 26 November 2006; accepted 26 November 2006

bstract

Many common activities such as walking in a shopping mall, moving in a busy subway station, or even avoiding opponents during sports,

ll require different levels of navigational skills. Obstacle circumvention is beginning to be understood across age groups, but studying trained

thletes with greater levels of motor ability will further our understanding of skilful adaptive locomotor behavior. The objective of this work

as to compare navigational skills during fast walking between elite athletes (e.g. soccer, field hockey, basketball) and aged-matched non-

thletes under different levels of environmental complexity in relation to obstacle configuration and visibility. The movements of eight women

thletes and eight women non-athletes were measured as they walked as fast as possible through different obstacle courses in both normal and

ow lighting conditions. Results showed that athletes, despite similar unobstructed maximal speeds to non-athletes, had faster walking times

uring the navigation of all obstructed environments. It appears that athletes can process visuo-spatial information faster since both groups can

ake appropriate navigational decisions, but athletes can navigate through complex, novel, environments at greater speeds. Athletes’ walking

imes were also more affected by the low lighting conditions suggesting that they normally scan the obstructed course farther ahead. This

tudy also uses new objective measures to assess functional locomotor capacity in order to discriminate individuals according to their level of

avigational ability. The evaluation paradigm and outcome measures developed may be applicable to the evaluation of skill level in athletic

raining and selection, as well as in gait rehabilitation following impairment.

2006 Elsevier B.V. All rights reserved.

www.elsevier.com/locate/gaitpost

Gait & Posture 26 (2007) 539–545

eywords: Gait; Obstacle avoidance; Locomotor control; Navigation; Elite athletes

1. Introduction

Humans are required to navigate around obstacles in

different situations such as when avoiding another pedes-

trian on the sidewalk, when walking in a busy shopping mall,

or during sporting events. The various complexities of these

environmental contexts require different navigational skills

implicating both the motor and cognitive aspects of the

control of locomotion. Normal human behavior during

* Corresponding author at: 525, boul. Wilfrid-Hamel, IRDPQ (CIRRIS),

uebec, Que., Canada G1M 2S8. Tel.: +1 418 529 9141x6584;

ax: +1 418 529 3548.

E-mail address: brad.mcfadyen@rea.ulaval.ca (B.J. McFadyen).

966-6362/$ – see front matter # 2006 Elsevier B.V. All rights reserved.

oi:10.1016/j.gaitpost.2006.11.209

obstacle circumvention is beginning to be understood in

children [1], young adults [2–5], and older adults [6,7].

Understanding how trained athletes view environmental

space and navigate will provide additional insights into

adaptive locomotor control.

Although many biomechanical studies have focused on

stepping over obstacles (e.g. Refs. [8,9]) the circumvention

of obstacles has received relatively little attention. Vallis and

McFadyen [5] recently showed that the motor behavior used

to circumvent an obstruction differs from when steering to

change one’s walking direction (e.g. Refs. [9,10]). Fajen and

Warren [2] proposed a dynamic model to predict naviga-

tional paths around virtual obstacles appearing at a given

distance while walking towards a goal. According to this

M. Gerin-Lajoie et al. / Gait & Posture 26 (2007) 539–545540

Table 1

Characteristics for the two test groups

Athletes (eight women) Non-athletes (eight women)

Mean S.D. Range Mean S.D. Range

Age (years) 21.1 1.0 20–23 21.6 2.9 18–28

Mass (kg) 63.1 6.9 55–73.5 69.3 9.9 59.5–88.3

Height (m) 1.70 0.09 1.63–1.89 1.72 0.06 1.65–1.84

model, walking trajectories are a function of the relative

angles between subject heading direction and the relative

positions of the final goal and obstacles acting as attractors

and repellers. The authors thus proposed that the route

through obstructed environments is not planned explicitly,

but rather emerges on-line from the walker’s interactions

with the environment.

Path planning has been proposed as another theory of

navigation in a cluttered environment. Patla et al. [3] argued

that considering only the next obstacle in the travel path is

not the strategy used to navigate a cluttered environment.

They provided some evidence that path planning is not based

on detailed geometrical information about the environment,

but rather on avoiding obstacle clusters in the line-of-sight

travel path. Gerin-Lajoie et al. [4] also showed that some

anticipation (center of mass deviations before the obstacle)

and some initial planning (constant clearance around the

obstacle) are involved in circumventing a human-shaped

obstacle.

With respect to athletes trained in obstacle avoidance

behavior, although knee mechanics during side-step cutting

maneuvers have been studied in novice and more

experienced athletes [11], no study, to our knowledge, has

investigated skilled navigation per se. Since motor expertise

is very task-specific [12], studying navigational behavior in

environments requiring greater demands in anticipatory and

sensorimotor control among athletes who are trained, due to

the nature of their sport, to avoid obstacles, should unveil

characteristics of skillful adaptive locomotor behavior used

to navigate a cluttered environment to reach a spatial goal.

Furthermore, developing measures and tests that can

discriminate individuals at different levels of anticipatory

navigational ability would be the first step towards the

elaboration of more detailed assessments of functional

locomotor capacity in various populations. Studying healthy

athletes will also provide a data base for comparison with

athletes who have sustained musculoskeletal injuries such as

an anterior-cruciate ligament rupture (a common injury

among women soccer and basketball players (e.g. Ref. [13]))

in order to study the effects of such injuries on navigation

strategies.

For the present study, we asked whether elite athletes,

who are involved in obstacle avoidance performance, have

better anticipatory navigational skills than non-athletes. If

so, what is it that they do better? Therefore, we compared

anticipatory navigational skills during fast walking between

elite athletes and aged-matched non-athletes under different

levels of environmental complexity relative to the number

and configuration of obstacles and the availability of visual

information due to lighting conditions. Athletes were

expected to navigate the obstacle courses in shorter times

when compared to non-athletes, especially with increasing

level of task difficulty. The assumption was that their

frequent rehearsal of obstacle circumvention maneuvers

would enable them to have better (space domain) and faster

(time domain) navigational strategies. In addition, given the

hypothesis that athletes scan the environment farther ahead

in order to better anticipate and plan an optimal route, the

low lighting condition was expected to have more impact on

this trained group.

2. Methods

2.1. Subjects

Eight elite women athletes from varsity or premier league teams

(five from soccer, one from basketball, one from field hockey, and

one from ultimate Frisbee) participated in the study. Such sports

were targeted because they involve frequent rehearsal of complex

navigational strategies during walking/running. Eight aged-

matched healthy non-athlete adult women (who participated in

sports, but on a recreational basis only) were recruited from the

University of Calgary Community. Subjects’ characteristics are

shown in Table 1. Ethics approval was obtained from the respective

university ethics boards and all participants provided written

informed consent. Exclusion criteria included: any self-reported

neurological or musculoskeletal problems; being shorter than 1.6 m

or taller than 1.9 m (to control for view of the obstacle course);

taking medications affecting alertness or locomotion; a score below

20/20 on the Snellen vision test (with corrective lenses when

necessary) and those as goaltenders for the athlete group. Elite

athletes of other sports or people involved in recreational versions

of the targeted sports more than once a month were excluded from

the control group. Female subjects were recruited for this study in

anticipation of follow-up work on the effect of an anterior-cruciate

ligament rupture on such navigational tasks.

2.2. Instrumentation

Three-dimensional movements of the subjects were tracked

(120 Hz) during walking using eight digital cameras (Eagle,

Motion Analysis Corp., USA) arranged to allow a top down view

of the markers minimizing marker occlusion. Three non-colinear

reflective markers were placed on the trunk and head segments.

Specific anatomical points (e.g. the humeral heads and sternal

notch for the trunk segment) were also digitized to allow the

estimation of segmental center of mass (CM).

Cylindrical obstacles were custom built and up to 14 were used

to create the different obstacle courses (Fig. 1A). Each of these

obstacles (height: 1.45 m, diameter: 0.3 m) was made of a fabric

shell filled with a stack of five inflated beach balls. Sandbags were

placed at the bottom of the obstacles to enable them to stand

upright. The walkway was delimited by a 4.6 m-wide by 10 m-long

carpet that had contrasting tape on its perimeter and a grid marked

on it to facilitate reproducible obstacle configurations. An auto-

matic, custom-built, curtain system (Fig. 1B) occluded vision of the

M. Gerin-Lajoie et al. / Gait & Posture 26 (2007) 539–545 541

Fig. 1. (A) Example of an obstacle course arranged over the 4.6 m-wide by 10 m-long walkway as viewed from gait initiation. Note the two posts forming a gate at

the end of the walkway through which subjects had to pass to complete the task. (B) Example of a subject at the starting position behind the retracting opaque curtain.

obstacle course before each trial. A string attached to the subject’s

waist and passing through a pulley mechanism retracted the

curtains as the subject stepped forward. A release mechanism

automatically detached the string from the subject just before

crossing the starting line. The curtain system provided uniform

exposure times to each obstacle course for all subjects. Two sets of

timing beams, aligned across both a starting and finishing line, were

used to measure and provide feedback about navigation time so as

to encourage subjects after each trial to keep motivation at a high

level. A set of posts set at the finish line to either side of the middle

of the walkway mimicked a 1.0 m-wide doorway that subjects had

to pass through to complete the task.

2.3. Protocol

Subjects first performed 15 practice trials including both unob-

structed and simple obstructed situations. The first six of these

familiarization trials were performed at normal walking speed,

while for the remaining trials, subjects were instructed to cover the

distance between the starting and finishing lines in the shortest

possible time. Subjects were allowed to walk as fast as they could,

without running (i.e. always at least one foot in contact with the

ground). Using fast versus normal walking speed created a more

challenging situation while providing a fair comparison because of

the expected speed ceiling effect across the two groups since

running was not allowed. Subjects were also instructed to avoid

contacting the obstacles and to remain within the delimited peri-

meter of the walkway at all times. Five observers dispersed around

the walkway ensured that subjects respected all instructions (the

rare faulty trials were repeated).

The test consisted of navigating, in the shortest possible time,

through three blocks of obstacle courses organized in a progression

of increasing number of obstacles and their proximity to the starting

line (Fig. 2). Each obstacle pattern was designed such that there was

an optimal path over which the subject would cover a minimal

amount of distance while conserving forward momentum. Each

block included three different obstacle courses that were performed

once in their original configuration, and once in a mirror image to

control for preferred side (since the first avoidance choice would

affect the performance for the remaining portion of the obstacle

course, this provided a fair comparison regardless of subjects’

lateral preference). The corresponding six trials were performed in

both normal lighting (florescent fixtures of the laboratory) as well

as in the low lighting condition (created by dimming the lights and

having subjects wear sunglasses) for a total of twelve trials per

block. The obstacle patterns and lighting conditions were presented

in a randomized fashion within each testing block.

2.4. Data analysis

Marker identification was performed using EVaRTTM (Motion

Analysis Corp., USA). Three-dimensional motion data were inter-

polated, filtered (low pass second order Butterworth filter, 6 Hz)

and dependent variables calculated using custom programs. The

trajectory of the CM was estimated from the respective CMs of the

trunk and head segments. The subject’s maximum walking speed

was taken from the shortest time of the unobstructed trials per-

formed in normal lighting conditions (average magnitude of the

CM velocity from the starting to the finish line). The time to

complete the task represented the travel time of the CM between the

starting and finish lines. Optimized path selection was calculated as

the number of times the optimal path was picked for a given

condition. Navigation efficiency was defined as the straight ahead

distance (10 m), divided by the actual CM traveled distance.

Relative speed was calculated as the average magnitude of the

actual CM velocity over the subject’s unobstructed maximum

walking speed. The time spent in the first 2-m unobstructed section

of the walkway was calculated as an indication of the time taken to

scan the environment at the beginning of locomotion.

Actual environmental difficulty was confirmed after data ana-

lyses based on the overall average time across subjects and con-

ditions to complete each obstacle course (where more time related

to greater difficulty). Statistical analyses and presentation of results

were then based on this determined difficulty order. For each

obstacle course, the mean between performance measures in the

original configuration and in its mirror image was used for analysis.

A t-test was used for the between-group comparisons of the

maximum unobstructed gait speed (significance levels set at

0.05 in all tests, SPSS, 11.0.0). ANOVAs were used to determine

main effects with regards to the difficulty block, subject group, and

lighting condition for the different dependent variables studied.

Where main effects were found, pairwise differences within each

difficulty block were sought between the two groups and the two

lighting conditions using t-tests.

M. Gerin-Lajoie et al. / Gait & Posture 26 (2007) 539–545542

Fig. 2. Blocks (each row) of obstacle courses as presented to subjects. Each obstacle pattern is illustrated in both its original (O) configuration and its mirror

configuration (M). The dotted lines represent the optimized path estimated in terms of minimum travel distance and the maximization of the conservation of

forward momentum. All of the illustrated obstacle patterns were presented in both normal and low lighting conditions.

3. Results

Maximum unobstructed walking speeds did not differ

between athletes (2.59 � 0.12 m/s) and non-athletes

(2.52 � 0.18 m/s). Time scores (Fig. 3A) are presented

according to the order of obstacle course difficulty that was

based on the global time to complete each obstacle course.

These scores therefore increase with the difficulty level.

Statistical analysis confirmed that the time scores were

significantly different between the respective difficulty

blocks (F2,276 = 54.41, p < .001). Athletes completed the

task in shorter times (F1,276 = 64.56, p < .001) for every

block of obstacle difficulty. The low lighting condition

resulted in slightly, but significantly, longer times for both

M. Gerin-Lajoie et al. / Gait & Posture 26 (2007) 539–545 543

Fig. 3. (A) Total time to complete the task, (B) optimized path selection, (C) navigation efficiency, (D) walking speed as a percentage of maximum values, and

(E) time spent in the first 2-m section of the walkway for athletes (A’s) and non-athletes (NA’s) during the normal and low lighting conditions. Data are arranged

in blocks of difficulty based on overall average walking time (see Section 2). Significant differences between groups as well as within group differences for the

lighting conditions are reported on the graphs ( p < .05). Error bars represent standard errors of the mean.

M. Gerin-Lajoie et al. / Gait & Posture 26 (2007) 539–545544

groups (F1,276 = 6.29, p = .013), but the only pairwise

significant difference for lighting conditions was found for

athletes in the most difficult block.

Optimal path selection (Fig. 3B) decreased with

increasing difficulty for both groups (F2,84 = 63.45,

p < .001). The related navigation efficiency (Fig. 3C) also

decreased as difficulty increased for both groups

(F2,276 = 111.43, p < .001). There were no main effects

for group and lighting conditions for either variable, with the

exception that athletes navigated, in general, slightly more

efficiently than non-athletes (F1,276 = 4.06, p = .045). None

of the pairwise comparisons between the two groups,

however, was significant.

Walking speed (Fig. 3D), expressed as a percentage of the

subject’s maximum walking speed, decreased with the

increasing level of difficulty in both groups (F2,276 = 44.00,

p < .001). Athletes, however, were able to navigate through

the obstacle courses at speeds closer to their own

unobstructed maximum walking speed (F1,276 = 33.76,

p < .001) in comparison to non-athletes. This was speci-

fically the case for blocks 2 and 3 where the pairwise

comparisons between the two groups all reached signifi-

cance. The low lighting condition forced both groups to

decrease their relative walking speed (F1,276 = 7.01,

p = .009).

The time spent in the first 2-m unobstructed section of the

walkway (Fig. 3E) increased with the difficulty level

(F2,276 = 25.82, p < .001). Athletes, however, spent in

general less time in that portion of the walkway

(F1,276 = 17.43, p < .001), with significant pairwise differ-

ences in the low lighting condition for block 1 and in all

conditions for block 2. The low lighting condition only

resulted in a tendency to increase the time spent in the first 2-

m section of the walkway.

4. Discussion

Overall, athletes have enhanced navigational skills

demonstrated by superior temporal scores across all

obstruction complexity levels in both normal and low

lighting conditions. Degrading visuo-spatial information

during the low lighting condition was more costly for

athletes in the most difficult obstacle course, suggesting that

athletes make better use of information further into the

obstacle course when it is available during navigation.

Two avenues are possible to explain better performance

times by athletes: the minimization of the traveled distance;

or the maximization of the traveling speed (conservation of

forward momentum). The first avenue relates to the medio-

lateral spatial strategies (path selection and navigation

efficiency). Athletes were expected to make better naviga-

tional decisions, but there was no significant difference in

the number of times that both groups selected the optimized

path (best tradeoff between minimizing distance and

maximizing speed). Interestingly, however, this trend is

reversed for the low lighting condition in block 3, again

suggesting that the lack of visual information is more costly

for the anticipatory strategies of athletes. Athletes were also

expected to do a better job at minimizing the total distance

traveled (navigation efficiency). This trend was observed,

but the gain is small and it does not appear to be sufficient to

explain the observed time performance differences.

The second possibility to explain group difference in

performance is to make a comparison in the time domain.

First, it is important to note that both groups had the same

unobstructed maximum walking speeds, meaning that any

difference in obstructed walking would thus have to be

explained by the ability to make better use of the available

speed. As the complexity of the obstructed environment

increased, athletes were able to navigate through the

obstacle courses at speeds closer to their maximum

unobstructed walking speed as compared to non-athletes.

The faster navigation speeds by athletes suggest that they

can process the visuo-spatial information faster than the non-

athletes.

A first trial approach without previous knowledge of the

next obstacle course was adopted to ensure that the motor

task posed a new situation to be resolved by the locomotor

system each time. Subjects, therefore, had to analyze the

obstacle course as they initiated walking as opposed to

employing a previously stored route. The time spent in the

first 2-m unobstructed section of the walkway was used as an

indication of the initial time taken to scan the environment

and to establish a navigation strategy. Results showed that

both groups had a similar level of success for selecting the

optimized path. Athletes, however, were capable of selecting

their path while spending, on average, less time in the first 2-

m unobstructed section of the walkway. This suggests that

athletes can more efficiently process the visuo-spatial

information about the upcoming path during navigation.

Yet, were the subjects in the present study actually

planning their path, or relying on a reactive type of control

based on online visual processing? The previously suggested

online visual information model [2] for navigation, in which

the observed route emerges from the interactions of the

observer with the next object in the near environment, is

attractive since it can quickly adjust to sudden changes. If

subjects in the present study would have been relying purely

on a reactive type of control based on the geometry of the

obstacle course at hand, then they would have always used

the mirror image of the selected path in the mirror image

presentation of the corresponding obstacle configuration.

Qualitatively looking at the individual paths taken, this was

not the case.

On the other hand, the explicit planning of a path by

grouping clusters of obstacles together and identifying safe

corridors [3] should require, in theory, more planning time

because of the level of information processing involved. If

the subjects of the present study were really explicitly

planning their path in advance while integrating information

about obstacles farther ahead, optimized path selection

M. Gerin-Lajoie et al. / Gait & Posture 26 (2007) 539–545 545

would have been higher than the 40% or 30% levels that

were, respectively, observed in both groups for the second

and third block. It should be noted, however, that using short

pylons (0.72 m high) as in Patla et al. [3] could favor the

described planning strategy in terms of identifying clusters

of obstacles. The present study used taller and wider

obstacles (1.45 m � 0.3 m) providing a different point of

view, which posed a greater challenge for scanning ahead.

So, are athletes planning or reacting faster? It seems

that they are not strictly relying on either a reactive or an

initial planning strategy to control navigation. It is clear

from the present results though that athletes can process

visual information (whether ahead of time or online) faster

than non-athletes of similar unobstructed walking ability.

In environments such as used here, it may be the case that

the locomotor control system uses both mechanisms, and

if switching between the two types of control during the

navigational task is the strategy that is used, athletes

would appear to be able to do it faster. In any case, more

research is needed to delineate environmental contexts

that elicit reactive and anticipatory visual control of

locomotion.

In conclusion, this study has shown that elite athletes,

trained in obstacle avoidance performance, have better

navigational skills in complex obstructed environments

when compared to non-athletes. The speed at which visuo-

spatial information can be processed seems to be the key

difference since both groups can make appropriate naviga-

tional decisions, but athletes can navigate complex, novel,

environments at speeds closer to their maximal walking

speed. It is suggested that the control of navigation involves

a mix of strategies between a reactive type of control and the

planning of a path. This study has also shown that it is

possible to use objective measures to assess complex

locomotor capacity and to discriminate individuals at

different levels in the motor capacity spectrum according

to their respective level of anticipatory navigational abilities.

This information is important because such measures of

navigational ability could be eventually exploited for talent

detection, athletic training, and rehabilitation programs.

Such quantitative assessment of locomotor capacity should

also lead to improved measures of recovery following injury

and, ultimately, to more objective indices of readiness for

return to play. Finally, the relationships between anticipatory

gait adjustments, the complexity of the environment and the

level of motor expertise still require more research to further

our understanding of pathological, normal, and highly

trained adaptive gait behavior.

Acknowledgements

This work was supported by an Interdisciplinary

Capacity Enhancement team grant from the Canadian

Institutes for Health Research (CIHR). Dr. Gerin-Lajoie was

also a Natural Sciences and Engineering Research Council

(NSERC) scholar. Dr. Janet Ronsky holds a CRC in

Biomedical Engineering funded by NSERC. The authors

would like to thank all of the people who participated in this

study. We also gratefully thank Matt Gotch, Angela Meier,

Jody Platt, Veronique Biron, Andrejz Stano, and Mr Guy St-

Vincent for their technical assistance.

References

[1] Vallis LA, McFadyen BJ. Children use different anticipatory control

strategies than adults to circumvent an obstacle in the travel path. Exp

Brain Res 2005;167:119–27.

[2] Fajen BR, Warren WH. Behavioral dynamics of steering, obstacle

avoidance, and route selection. J Exp Psychol Hum Percept Perform

2003;29:343–62.

[3] Patla AE, Tomescu SS, Ishac MG. What visual information is used for

navigation around obstacles in a cluttered environment? Can J Physiol

Pharmacol 2004;82:682–92.

[4] Gerin-Lajoie M, Richards CL, McFadyen BJ. The negotiation of

stationary and moving obstructions during walking: anticipatory

locomotor adaptations and preservation of personal space. Motor

Control 2005;9:242–69.

[5] Vallis LA, McFadyen BJ. Locomotor adjustments for circumvention

of an obstacle in the travel path. Exp Brain Res 2003;152:409–14.

[6] Gerin-Lajoie M, Richards CL, McFadyen BJ. The circumvention of

obstacles during walking in different environmental contexts: A

comparison between older and younger adults. Gait Posture

2006;24:364–9.

[7] Reed RJ, Lowrey CR, Vallis LA. Middle-old and old-old retirement

dwelling adults respond differently to locomotor challenges in clut-

tered environments. Gait Posture 2006;23:486–91.

[8] McFadyen BJ, Winter DA. Anticipatory locomotor adjustments during

obstructed human walking. Neurosci Res Commun 1991;9:37–44.

[9] Patla AE, Prentice SD, Robinson C, Neufeld J. Visual control of

locomotion: strategies for changing direction and for going over

obstacles. J Exp Psychol Hum Percept Perform 1991;17:603–34.

[10] Grasso R, Prevost P, Ivanenko YP, Berthoz A. Eye-head coordination

for the steering of locomotion in humans: an anticipatory synergy.

Neurosci Lett 1998;253:115–8.

[11] Sigward S, Powers CM. The influence of experience on knee

mechanics during side-step cutting in females. Clin Biomech

2006;21:740–7.

[12] Abernethy B. Training the visual-perceptual skills of athletes. Am J

Sports Med 1996;S-89:92.

[13] Micheli LJ, Metzl JD, Di Canzio J, Zurakowski D. Anterior cruciate

ligament reconstructive surgery in adolescent soccer and basketball

players. Clin J Sport Med 1999;9:138–41.