Content uploaded by Pooya Soltani
Author content
All content in this area was uploaded by Pooya Soltani on Jun 26, 2015
Content may be subject to copyright.
Proceeding of 7th European Conference on Games
Based Learning (ECGBL2013)
Porto, Portugal
793
Exploring learning effects during virtual swimming using biomechanical analysis (a work in progress)
Pooya Soltani, João Paulo Vilas-Boas
Porto Biomechanics Laboratory (LABIOMEP), Faculty of Sport, University of Porto, Porto - Portugal
111114024@fade.up.pt
jpvb@fade.up.pt
CITATION (APA):
Soltani, P., & Vilas-Boas, J. P. (2013). Exploring learning effects during virtual swimming using
biomechanical analysis (a work in progress). In P. Escudeiro, & C.V. de Cravalho. (Eds.). Paper presented
at Proceedings of the 7th European Conference on Games Based Learning (Portugal), Porto (pp. 793-
796). Sonning Common, UK: Academic Conferences and Publishing International.
Abstract: In this work-in-progress (WIP), kinematics of movement during playing virtual swimming
exergame between novice and experienced players was analyzed in order to detect possible learning
effects. Ten participants performed a 100-meter front crawl virtual swimming using Xbox and Kinect. A
12 camera Qualisys motion capture system tracked the position of 22 reflective markers simultaneously.
Preliminary results indicate that novice players showed a pattern closer to real swimming, spending
more time on accuracy of movement, whereas the experienced players used an adaptive circular
pattern in order to win the game.
Keywords: Exergame; Virtual Sport; Learning; Biomechanics
Introduction: Since 2000, new types of video game consoles (i.e. Microsoft Xbox and Kinect, Nintendo
Wii or Sony Move) often tagged as Exergames have been introduced in which users have to interact
physically in order to advance in the game. Claiming that Exergames can promote an active lifestyle,
they don’t actually increase levels of physical activity but rather the motivation to perform some basic
repetitive movements (Peng et al., 2013).
As one of the genres of popular games, virtual sports encourage participants to perform the activity as
though they are playing the real sport. These systems use several methods to detect the movements of
players. Adaptations to active video games may happen after playing for some time. In fact, most of the
users start playing technically rather than emotionally which means that they will learn how to play the
game with less effort (Pasch et al., 2009). One of the causes is related to specifications of the platform
used, which allows for the use of such strategies. For example, Nintendo Wii uses a sensor which detects
the movement with an internal accelerometer. Although the optimum goal is to perform the complete
movements in virtual sports, there is always the possibility of “cheating” due to the user being able to
easily fool the accelerometer by using one’s wrist to manipulate the controller (Lavac et al., 2009). This
possibility is often discovered through experience or learning other players’ styles of playing.
On the other hand, as there is no exact force confronting players’ actions, there are some concerns
regarding efficacy and safety during the use of Exergames, especially when high exposure to the game is
expected. Characterizing Exergames biomechanically allows designers to create more realistic games for
Proceeding of 7th European Conference on Games
Based Learning (ECGBL2013)
Porto, Portugal
794
a more meaningful experience; it will allow hardware developers to identify the limitations of their
platforms in order to minimize the effects of learning; reduce musculoskeletal injuries (accidents due to
limitations of players both in terms of space and muscle activity), and create additional resources (e.g.
warm up and cool down periods).
Therefore the purpose of this work-in-progress is to compare spatiotemporal parameters between
novice and experienced players to characterize possible learning occurring during playing virtual sports.
Methods: Ten healthy subjects (6 male and 4 female; Age: 22-31) played Michael Phelps: Push the Limit
(505 Games, Milan, Italy) virtual swimming game using Xbox 360 and Kinect (Microsoft, Redmond, WA).
Subjects were divided into two groups of “Bad Performers” if their ranking during the game was 1st to 4th
and “Good Performers” if their position was 5th to 8th. All testing was conducted on the dominant arm.
Virtual Swimming: Subjects had to stand in front of the Kinect sensor and had to bend forward and as
soon as they saw the “Go!” command, they had to return back to normal position with arms in front.
After that, they had to swing their arms in order to move the avatar in the game (100-meter front crawl
swimming). At the end of the event, they had to drop both hands and then raise one of them to finish
the race. In order to prevent the player from swimming too fast or too slow, there is a spectrum on the
screen with a blue zone in between which indicates if the speed is at the moderate level. At the middle
of the second lap, there is a possibility to swim as fast as possible called “Push the Limit”. We called it
“fast” swimming.
Collection of Kinematics data: Dominant arm’s movements were recorded at 200 Hz with a 12-camera
Qualisys motion capture system (Qualisys AB, Gothenburg, Sweden). Before each experiment, the
cameras were calibrated to the measurement volume. 22 spherical reflective markers were placed over
the skin. Collection of data was started when the subjects were placed at the “ready” position over a
period of one minute. The data were collected in acquisition software (Qualisys Track Manager).
Swimming technique and definition of cycles: The event was divided into four phases as suggested by
Maglischo (1993). Based on the speed (Normal Vs. Fast), Good and Bad Performers were compared. The
swimming cycle began with stretching the right arm forward which is considered as “Entry”. As the so
called propulsive phase of left arm finished, a phase called “Downsweep+Catch” starts. Some
characteristics of this phase are different than in real swimming due to the position of the body.
Following this phase, the hand moves in a circular sweep known as the “Insweep”. Hand direction
switches out, back, and up in a term called “Upseep”. This phase finishes as the hand passes the thigh
where its direction changes from back and up, to up and forward which is known as the “Recovery”
phases.
Statistical Analysis: All data was statistically analysed using SPSS version 20. An independent sample t-
test was conducted to compare spatiotemporal parameters in two performance groups. Statistical
significance was accepted at p ≤ 0.05.
Results: Total time to complete the event, numbers of cycles, start, average and max velocities, and
angular distance covered during the event are presented in Table 1. A sample movement pattern
created by a marker placed on the dominant hand in line with the index finger for both performance
groups was shown in Figure 1. As shown especially in superior view, bad performers’ moving pattern is
Proceeding of 7th European Conference on Games
Based Learning (ECGBL2013)
Porto, Portugal
795
closer to real swimming following swimming phases in their performance while the good performers
were creating a circular pattern. Time Motion Analysis and percentage of time dedicated to each phase
is shown in Table 2.
As shown in table 1, Good performers completed the event in a significantly shorter time (t(4.45)=-4.99,
p = 0.01). They also had lower numbers of cycles during both normal and fast swimming (t(8)=-2.64, p =
0.03 and t(4.34)=-3.03, p = 0.03 respectively). There were no differences in start velocity between the
two groups (t(8)=-0.19, p = 0.86). Good performers had significantly lower velocity during normal phase
of swimming (t(8)=-2.24, p = 0.05). However, there were no significant differences in average velocity
and maximum velocity during fast swimming phase (t(8)=-2.95, p = 0.78 and t(8)=-1.34, p = 0.22
respectively). Bad performers covered significantly more angular distance (t(8)=-2.84, p = 0.02).
Table 1: Spatiotemporal parameters of 100-meters front crawl virtual swimming in two levels of
performance
PARAMETERS
PERFORMANCE
N
MEAN
STD. DEVIATION
TOTAL TIME OF EVENT* (s)
GOOD
5
48.40
0.89
BAD
5
57.00
3.74
TOTAL NUMBERS OF CYCLES*: NORMAL
GOOD
5
36.60
3.21
BAD
5
46.40
7.63
TOTAL NUMBERS OF CYCLES*: FAST
GOOD
5
8.80
0.45
BAD
5
11.80
2.17
START VELOCITY (m.s-1)
GOOD
5
5.0640
1.64
BAD
5
4.8760
1.53
AVERAGE VELOCITY*: NORMAL (m.s-1)
GOOD
5
2.5160
0.52
BAD
5
3.1720
0.39
AVERAGE VELOCITY: FAST (m.s-1)
GOOD
5
3.9660
0.91
BAD P
5
4.1240
0.78
MAX VELOCITY: FAST (m.s-1)
GOOD
5
5.9100
1.10
BAD
5
7.1480
1.74
TOTAL DISTANCE COVERED* (m)
GOOD
5
115.00
24.32
BAD
5
160.60
26.35
*: Significant differences were observed between the two performance groups
Proceeding of 7th European Conference on Games
Based Learning (ECGBL2013)
Porto, Portugal
796
FRONTAL VIEW
SAGITAL VIEW
SUPERIOR VIEW
BAD
PERFORMERS
GOOD
PERFORMERS
Figure 1: Sample hand pattern in one complete cycle in front crawl
Table 2: Time Motion Analysis of different phases of virtual swimming between novice and expert
players
FRONT CRAWL
PERCENTAGE OF TIME IN EACH PHASE: NORMAL [FAST SWIMMING]
ENTRY
DOWN+CATCH+INSWEEP
UPSWEEP
RECOVERY
GOOD PERFORMERS
25±5 [25±1]
28±5 [25±3]
19±7 [20±2]
26±3 [27±2]
BAD PERFORMERS
30±7 [30±6]
29±4 [27±4]
16±2 [19±3]
21±5 [22±5]
Numbers are expressed in Mean±SD.
Discussion: The purpose of this WIP was to detect the learning effects occurring during the playing of
virtual sports using biomechanics. The preliminary results show that, in general, the movement pattern
of bad performers was closer to real swimming. As a result of learning, good performers were focused
more on completion of the circular movement.
Bad performers wanted to adapt their movements to the spectrum and sometimes they swung their
arms slower in order to decrease the speed of the avatar in the game. They dedicated more time to
learning how to adapt to the game. Average velocity was lower in good performers because of the
consistency of their movements. As good performers had lower numbers of cycles, angular distance
covered was also lower. The good performers dedicated more time to the recovery phase as they were
creating a circular pattern while bad performers spent more time to the down+catch+insweep phase
which shows that they were following the real swimming pattern and therefore spent more time in this
phase.
Proceeding of 7th European Conference on Games
Based Learning (ECGBL2013)
Porto, Portugal
797
According to our observations, as bad performers entered the fast swimming phase (Push the Limit),
they stopped following the real-swimming-movements. Good performers prioritized their movements to
make more circular pathways. As good performers learned how to play the game, they perceived that
real-swimming-movements were not necessary to play the game.
Previously, virtual reality has been used to teach in several training programs (Snyder et al., 2011). As
these gaming environments provide similar environments (Miles et al., 2012), and because of time
constraints, virtual sport exergames might be a good alternative to be considered in tandem with
traditional training methods. The biomechanical analysis used in the games may provide an image of
how reliable these systems are in terms of teaching and structuring practice; especially when it comes to
unsupervised long term usage.
Exergames may benefit players physiologically, but since the possibility of cheating exists due to the
technical limitations of different platforms, developing a detailed characterization of these games may
provide valuable information to be used in the feedback provided by these games. If we consider the
process of learning during the use of exergames, good performers adapted their movement in order to
win the game while bad performers were more focused on the accuracy of their movements. It seems
that the Kinect sensor is not able to detect the delicate movements which may prove to be a good
reason to switch from performing real movement to adaptive movements in exergames.
References
Levac, Danielle, Pierrynowski, Michael R., Canestraro, Melissa, Gurr, Lindsay, Leonard, Laurean, &
Neeley, Christyann. (2010) "Exploring children’s movement characteristics during virtual reality
video game play" Human Movement Science, Vol. 29, No. 6, pp 1023-1038.
Maglischo, E. W. (1993) Swimming even faster. Mayfield Publishing Co., Mountain View, Calif.; United
States.
Miles, Helen C., Pop, Serban R., Watt, Simon J., Lawrence, Gavin P., & John, Nigel W. (2012) "A review of
virtual environments for training in ball sports" Computers & Graphics, Vol. 36, No. 6, pp 714-
726.
Pasch, Marco, Bianchi-Berthouze, Nadia, van Dijk, Betsy, & Nijholt, Anton. (2009) "Movement-based
sports video games: Investigating motivation and gaming experience", Entertainment
Computing, Vol. 1, No. 2, pp 49-61.
Peng, Wei, Crouse, Julia C., & Lin, Jih-Hsuan. (2013) "Using Active Video Games for Physical Activity
Promotion: A Systematic Review of the Current State of Research", Health Education &
Behavior, Vol. 40, No. 2, pp 171-192.
Snyder, Christopher W., Vandromme, Marianne J., Tyra, Sharon L., Porterfield, John R., Jr., Clements,
Ronald H., & Hawn, Mary T. (2011) " Effects of virtual reality simulator training method and
observational learning on surgical performance", World Journal Of Surgery, Vol. 35, No. 2, pp
245-252.
