Application of tactile displays in sports: where to, how and
when to move
J a n B . F. v a n E r p 1 Ian Saturday 2 Chris Jansen 3
TNO Human Factors Utrecht University TNO Human Factors
ABSTRACT rowing motion feedback system as compared to traditional
In this paper we explore the possibilities of tactile displays in
sports applications, and report an experiment that shows that a
tactile feedback systems improves rowing efficiency compared
to traditional feedback systems. Earlier papers have shown that
localized vibrations provide intuitive cues for orientation and
navigation, i.e. where to move to, and motion initiation, i.e. how
to move. In the first part of the paper we will give examples for
the spin-off of these applications of tactile displays to the sports
domain, including tactical guidance for soccer players (where)
and body posture feedback for speed skaters and cyclists (how).
In part two we report a study that extends the ‘where’ and ‘how’
examples with coordinated movement patterns. These systems
also provide cues on when to move. In a laboratory experiment
we showed that motion feedback with a localized and timed
tactile cue resulted in better performance than the current
methods of motion feedback.
Keywords: tactile, motor behavior, sport, motion feedback,
Challenging environments such as those encountered by Figure 1: Helicopter pilot showing the TNO Tactile Torso Display designed
for orientation and navigation in challenging environments but with a
athletes, astronauts, pilots or individuals with a disability are a possible spin-off to sports.
major thrust for ergonomic innovation. An example of this
ergonomic thrust is the introduction of tactile torso displays to
counteract the danger of visual and auditory overload for
military pilots. Tactile displays use the skin as an information
channel. A widespread example is the vibration function on
mobile phones. Tactile torso displays consist not of one, but of a
matrix of vibrating elements and use a localized vibration on the
torso to present spatial information. Research has shown that
these displays can intuitively present spatial information, like
the proverbial tap on the shoulder: left is left, front is front, etc
. These displays have shown their value in, for example,
counteracting spatial disorientation , waypoint navigation
[15, 16] and low-level flight , see Figure 1 for an example.
Besides orientation and navigation information (i.e.,
information on where to move to), a localized vibration can also
be used to initiate limb movement (i.e., information on how to
move). This means that tactile displays could be used to learn
new or to fine tune existing co-ordination patterns . Figure 2
gives an example in which three people perform a coordinated
dance based on tactile signals given on their individual tactile
Both application areas are relevant for sport as well. Over
the past few years a spin-off of (military) tactile technology to
the sports domain is emerging. In the first part of this paper we
will describe several chances for tactile technology and describe Figure 2: Three ‘dancers’ showing the concept of multi-player coordinated
examples in which top coaches and sportsmen and women movement based on signals displayed on each participant’s tactile body
explored the technology. In part two we will report a controlled display. Picture taken during the Worldhaptics conference 2005 in Pisa.
laboratory study in which we investigated the effects of a tactile
Zijlaard-van Moorsel (see Figure 5) has been involved in
2 TACTILE DISPLAYS IN SPORT testing innovative posture feedback systems for cyclist in a
wind tunnel. For cyclists, the idea is that a tactile system could
Recently several tactile display applications have been improve seating and more importantly back position which is
investigated in the field of sport. We distinguish three classes: a. critical in maintaining a good aerodynamic posture.
tactical information in field sports, b. posture information, and c.
motion co-ordination pattern, that is where to, how and when,
An application that can be considered a spin-off of the
orientation and navigation displays is a system developed for
soccer players (see Figure 3). This system provides the soccer
player with tactical information on where to look and where to
move to through vibrators at specific locations on the body. One
of the problems that coaches experience during training is the
difficulty of communicating that kind of information verbally in
a timely and efficient manner, a. because of the distance
between coach and player, and b. because of possible language
barriers. Currently, top coaches may easily have five or more
languages spoken in a single team. The system uses the same
principles that are so successfully tested in mobility systems. A
vibration in a belt around the torso indicates the direction to
move to. This set-up is extended with four other signals: a Figure 4: Olympic champion speed skater training with position
vibration high up the back means “keep your head up”, one on measurement and feedback system.
the left shoulder means “look left” the one on the right shoulder
means “look right”, and one at the middle of the chest means
“stop”. By a wireless link, the coach can give an individual
player direct and language-barrier-free directions. The system
was tested by soccer coach Guus Hiddink while coaching
Champions League team PSV.
Figure 5: Olympic champion cyclist during a wind tunnel session to
optimize her back posture.
One step beyond posture is movement coordination [10,
22] or simply put a combination of posture and timing. The
sequence of limb motions and the forces exerted is critical in
many sports, including golfing, throwing the javelin, rowing,
Figure 3: The PSV soccer team evaluating position measurement and tennis etc. For these applications, not only motion initiation is
feedback systems. critical (how to move), but also the exact timing (when to
move). An example that will be used in the second part of
The second application is related to movement initiation this paper is rowing. Although rowing seems a relatively
applications. Body posture is a very critical aspect in virtually simple, cyclic motion, the timing of the different phases of the
all forms of sports, for instance in darts, golf, gymnastics, rowing motion has large effects on the force exerted on the
skiing, shooting, skating, dancing, etc. Important in the training oars [1, 2].
of posture is watching an expert and yourself taking a specific
posture. However, most coaches will also physically push and 3 TACTILE MOTION FEEDBACK IN ROWING
pull their student’s limbs and other body parts until the posture
is to his or her satisfaction. However, in a dynamic situation
(e.g. in speed skating or cycling) this is impossible. It is not so 3.1 Introduction
difficult to imagine a situation in which the hand of the coach is
The success factors of sports training have been extensively
replaced by a vibrating element. This concept was introduced in
studied. The four factors mentioned by Thorndike almost 80
training the posture of speed skaters (see Figure 4). Especially
years ago  are generally still accepted [4, 8, 11, 19],
for long distances, skaters tend to neglect their posture because
although debated by some researchers [9, 20, 21]. These four
of tiredness. An important “error” is that the skater raises his or
factors are: a. frequency of feedback (more is better), b. the
her shoulder, which has large negative effects on performance.
accuracy (higher is better), the timeliness (sooner is better),
As a spin-off of the movement initiations studies, a system was
and d. the information richness. Looking at the characteristics
developed that provides corrective instructions to the skater
of a tactile feedback system, it provides cues on when and
based on an advanced posture measuring system and a wireless
how to make a specific motion. Although Thorndike did not
link to the coach. Several Olympic skaters recently tested this
explicitly include cues on how to move as a success factor, we
system. Olympic and world champion cycling Leontien
argue that this is also a potential advantage of a tactile feedback right one for the back (see Figure 7, left panel). The only
system. The design of the tactile feedback system must be such difference with Group I was the location and modality of the
that it explores these potential advantages. information. Group III received delayed, non-positional
We studied these potential advantages in a rowing feedback. This group also received feedback on each stroke,
experiment. We choose rowing because it is cyclic and not too but only once every 25 strokes (see Figure 7, right panel). This
fast of a movement pattern while requiring accurate timing of latter condition mimics a human coach, although a human
the different phases. Also, the current indoor rowing machines coach would probably summarize the 25 strokes in one
allow to run the experiment in a controlled lab setting and to instruction.
automatically register several performance measures. Without The apparatus (see also Figure 8) consisted of a Concept 2
going into much detail, we will give a short description of model D indoor rowing machine. The built-in performance
rowing motion (see [3, 7] for more information). The critical monitor only depicted the remaining time of a session and the
part of the coordinated motion is the relation between the legs tempo (number of strokes per minute). Three Xsense MT9-B
and the back (or the knee angle and back angle), both in the pull motion trackers were used to measure the knee and back
phase (going backwards) as in the recovery phase (going angles. The system calculated both angles at 100 Hz. Figure 8
forward). An optimal rower will start with stretching the knees depicts the most relevant parts of the set up, including the
with a steady back until a specific knee angle is reached and the rowing machine, the motion sensors and the control panel of
knees and back work together (see Figure 6). the experiment leader. The tactile feedback was provided by
pager motor based tactile actuators (TNO model JHJ-3). These
actuators have a contact area of 1 by 2 cm and vibrate at
approximately 160 Hz. Heart rate of the participants was
measured with a Polar device.
Figure 6: Sketch used by trainers to explain the (sequential) relation between
the acceleration of the legs, trunk, and arms in rowing.
The two most common faults are moving the back either too
early or too late. In a standard feedback situation a rowing coach Figure 7: Feedback display for group II with flashing circles for the knees
will give verbal feedback (e.g., “you are too late with your and back (left) and group III with an overview of the back timing after 25
strokes (right, figure truncated after 15 strokes).
back”) every ten or so rowing motions. A tactile feedback
system could provide this feedback during (and not after) every
pull and recovery. Moreover, making the feedback positional, it
could also indicate how to move. We call this direct and
positional, respectively. To assess the effect of these two
potential success factors we compared the training results of
three groups: group I with direct and positional feedback via
vibrators on knees and back, group II with direct but non-
positional feedback via a visual display, and group III with
delayed (after 25 strokes), non-positional feedback. The
feedback was held constant with respect to the other success
factors accuracy, frequency, and information richness.
A total of 22 female rowers participated in the experiment with a
mean age of 20.1 year (SD 1.06). All participants trained at Figure 8: Participant in the rowing machine and with the motion sensors
attached to the knees and back. The monitor in front allowed the
student league level and had at least two supervised training experiment leader to check the motions and other parameters.
sessions each week. The participants were randomly assigned to
one of the three experimental groups. Each participant rowed eight sessions on one day: a pre-
The three groups differed in the way feedback was given. test, six training sessions and a post-test. Feedback was only
However, all three groups received feedback on whether and given during the six training sessions. Each session was ten
how much their back was too early or too late in the pull phase minutes in length with a 20-minute break in between sessions,
of each stroke. Thus all three forms of feedback had the same and a 1-hour lunch break after the fourth session. The
accuracy, frequency and information richness. Group I received participants were instructed to maintain tempo of 21 strokes
direct positional feedback. This was provided by two sets of per minute which they could check on the display of the
vibrators, one on the knees and one on the back. These vibrators rowing machine. They were also instructed to maintain a heart
gave a burst of vibration when a set knee and when a set back rate of around 160 beats per second. The experiment was run
angle was reached, respectively. A correct coordination would in the presence of their own experienced and certified rowing
result in simultaneous vibrations while the situation of back too coach. The starting point for the feedback was an 80° knee
early / too late would result in the back vibrators signaling angle and a 110° back angle, but individual adjustments of
before / after the knee vibrators. Group II received direct, non- these angles could be made by the coach. However, once the
positional feedback. For this group, the vibrators were replaced values were chosen, they were used throughout the
with two circles on a monitor, the left one for the knees and the experiment.
3.3 Results pattern, potentially even during each stroke (for the direct,
non-positional group). Although high performance levels can
We calculated the following dependent variables: timing (the
be maintained through these (continuous) corrections, it comes
absolute difference between reaching the set knee angle and the
at a price, that is it requires more effort which is reflected in
set back angle in ms), heart rate (beats/minute) and speed
an increased heart rate over the day. The second is that the
(km/hour). All data were analyzed with an analysis of variance
increased heart rate reflects higher mental effort in the groups
(ANOVA): test (pre-test, post-test) × feedback group (direct with non-positional feedback . This explanation would be
positional, direct non-positional, delayed, non-positional). The in line with the findings with other tactile displays that apply
timing data showed no significant results. Table 1 gives the an intuitive, localized cue and that reduced the mental
means of all groups and sessions. workload of the user. Concluding, we can say that the very
high performance level can be maintained with significantly
Table 1: Timing accuracy (ms) as function of feedback condition and session.
tr means training session.
lower effort when trained with tactile feedback.
Feedback In this paper we gave examples of the possible use of tactile
Group displays in the sports domain. Proof-of-concepts were
129 89 90 88 83 82 87 85 described of a tactical display (or where to move to) for soccer
II: direct, non- players, a posture display (or how to move) for speed skaters
83 93 71 53 60 64 52 79 and cyclists, and a movement coordination display (or when to
III: delayed, move) for rowers. In the lab experiment on a rowing machine,
78 75 72 73 60 56 79 84
non-positional we implemented a tactile feedback system to train the timing
of the limbs involved in rowing. In rowing, the efficiency of
The ANOVA of the speed resulted in no significant effects. the action is determined by the ability to time leg, trunk, and
But the ANOVA of the heart rate showed a significant test × arms movements. An important aspect in this sense is the
group interaction; F(2, 16) = 5.14; p < .02. The means are relation between the knee angle and back angle as function of
depicted in Figure 9. stroke phase.
Although both coach and trainees liked the system, future
studies have to confirm the potential of localized tactile
feedback in learning and improving motion patterns. When
heart rate (beats /minute)
168 people can execute a movement pattern guided by tactile cues
166 alone, the possibilities in the sports and in other domains are
sheer endless. For example, motion sensors could register the
movement pattern of an expert which can be replayed on the
tactile guidance system of the student.
160 In general, the applications are based on relatively simple
I: direct, positional
158 II: direct, non-positional technolgy and concepts but have proven to be quite effective.
III: delayed, non-positional
They have shown to be able to solve pratical problems such as
pre post language barriers and longer distances and to be able to
test provide direct and intuitive feedback. The displays are not
only hands, eye, and ears free, but also mind free.
Figure 9: Heart rate differences between the pre- and post-test for the three We conclude that the recent developments in the field of
experimental groups. tactile displays driven by challenging situations (e.g., pilots
and astronauts) have a potentially beneficial spin-off to sports
3.4 Discussion applications.
None of the three groups showed either a positive or negative
learning effect on timing or speed, although a positive effect was REFERENCES
expected for all three groups. The feedback given complied with  Baudouin, A., & Hawkins, D. (2002). A biomechanical review of
at least three of the success factors for motion feedback in the factors affecting rowing performance. Journal of Sports
sense that the feedback was accurate, frequent (feedback on each Medicine, 36, 396-402.
individual stroke), and information rich. For groups I and II it  Baudouin, A. & Hawkins, D. (2004). Investigation of
was also timely, and for group one even positional. The most biomechanical factors affecting rowing performance. Journal of
Biomechanics, 37, 959-976.
probable cause for the lack of direct performance improvement
 FISA competitive commission, (2002). Be a coach: Handbook –
might have been a ceiling effect. All rowers were highly Level 1. firstname.lastname@example.org
motivated to participate and anxious to perform well under the  Gauthier, G.M. (1984) Visually and acoustically augmented
eyes of their coach so that there was little room for improvement performance feedback as an aid in motor control learning: a
left. This ceiling effect leaves open the question whether direct, study of selected components of the rowing action. Journal of
positional feedback can improve movement coordination or not. sports sciences, 3, 3-26.
Interestingly, there is a substantial effect of feedback condition  Glencross, D.J. (1973). Temporal organization in a repetitive
on the heart rate. The group with tactile (i.e., direct and speed skill. Ergonomics, 16, 765-776.
 Jansen, C., Oving, A. & Van Veen , H.A.H.C. (2004). Vibrotactile
positional) feedback showed a decreased heart rate in the post-
movement initiation. Proceedings of Eurohaptics 2004, pp. 110-
test as compared to the pre-test, while the other two groups 117. München, Germany: Technische Universität.
showed a heart rate increase. This means that the latter two  Kleshnev, V. (2004). Propulsive efficiency of rowing. Ref Type:
groups had to invest more effort during the day to maintain their Internet communication;
optimal performance. Two factors may have played a role. The http://www.coachesinfo.com/category/rowing/77
first is the physical factor. When movement coordination is less  Newell, K.M. & McGinnis, P.M. (1985). Kinematic information
stable, the participant will be making small adjustments in the feedback for skilled performance. Human Learning, 4, 39-56.
 Salmoni, A.W., Schmidt, R.A., & Walter, C.B. (1984). Knowledge
of results and motor learning: A review and critical reappraisal.
Psychological Bulletin, 95, 355-386.
 Schmidt, R.A., & Lee, T.D. (2002). Motor Control and Learning: A
behavioral emphasis. (3 ed.) United States of America: Human
 Schmidt, R.A., & Young D.E. (1991). Methodology for Motor
learning: A paradigm for kinematic feedback. Journal of motor
behavior, 23, 13-24.
 Thorndike, E.L. (1927). The law of effect. American Journal of
Psychology, 39, 212-222.
 Van Erp, J.B.F. (2005). Presenting Directions with a Vibro-Tactile
Torso Display. Ergonomics, 48, 302-313.
 Van Erp, J.B.F., Groen, E.L., Bos, J.E. & Van Veen, H.A.H.C.
(2006). A Tactile Cockpit Instrument Supports the Control of Self-
Motion During Spatial Disorientation. Accepted by Human Factors.
 Van Erp, J.B.F. & Van Veen, H.A.H.C. (2004). Vibrotactile in-
vehicle navigation system. Transportation Research part F: Human
Factors, 7, 247-256.
 Van Erp, J.B.F., Van Veen, H.A.H.C., Jansen, C. & Dobbins, T.
(2005). Waypoint Navigation with a Vibrotactile Waist Belt.
Transactions on Applied Perception, 2 (2), 106-117.
 Van Erp, J.B.F., Veltman, J.A. & Van Veen, H.A.H.C. (2003). A
tactile cockpit instrument to support altitude control. Proceedings
of the Human Factors and Ergonomics Society 47th Annual
Meeting, pp. 114-118.
 Veltman JA, Gaillard AW. (1998). Physiological workload
reactions to increasing levels of task difficulty. Ergonomics, 41(5).,
 Wulf, G., & Schmidt, R.A. (1989). The learning of generalized
motor programs: reducing the relative frequency of knowledge of
results enhances memory. Journal of experimental psychology:
Learning, Memory and cognition, 15, 748-757.
 Wulf, G., Schmidt, R.A., & Deubel, H. (1993). Reduced feedback
frequency enhances generalized motor program, learning but not
parameterization learning. Journal of experimental psychology:
Learning, Memory and cognition, 19, 1134-1150.
 Wulf, G., & Weigelt, C. (1997). Instructions about physical
principles in learning a complex motor motor skill: to tell or not to
tell. Reseacrh quarterly for exercise and sport, 68, 362-367.
 Zanone, P.G., & Kelso, J.A.S. (1992). Evolution of behavioral
attractors with learning: Nonequilibrium phase. Journal of
Experimental Psychology: Human Perception and Performance, 18,