=Paper=
{{Paper
|id=Vol-1782/paper_9
|storemode=property
|title=Proposing of Planning System for Sports Domain: A Tool for Professional Coaches
|pdfUrl=https://ceur-ws.org/Vol-1782/paper_9.pdf
|volume=Vol-1782
|authors=Tomáš Škeřík,Wolfgang Faber,Lukáš Chrpa
|dblpUrl=https://dblp.org/rec/conf/plansig/SkerikFC16
}}
==Proposing of Planning System for Sports Domain: A Tool for Professional Coaches==
https://ceur-ws.org/Vol-1782/paper_9.pdf
Proposing of Planning System for Sports Domain:
A Tool for Professional Coaches
Tomáš Škeřı́k, Wolfgang Faber, Lukáš Chrpa
School of Computing and Engineering
The University of Huddersfield, Queensgate
Huddersfield, HD1 3DH
Abstract Sports Domain
In the 21st century, computing and network technologies are
This paper introduces the System for Sports Training daily life necessities and also play more and more signifi-
Planning (SSTP), which aims to be used as an auto- cant roles in sports training. The most significant progress
mated planning application that generates training plans has been made in the usage of computing in scientific sports
for individual athletes. This paper mainly presents the
sports domain, the system’s high-level architecture, on-
training, which leads to a systematic approach of train-
going development, and its challenges. ing, performance measurement, and competition analysis.
This contributed to training efficiency improvement, athletes
performance and it also helped to prevent some sports in-
juries (Hou 2015). Computing technologies involved in the
Introduction sports domain usually are data acquisition and data pro-
cessing, databases, modelling and simulation, which are
Recently, athletes performance developed beyond all expec- used in training and coaching, biomechanics, sports equip-
tation and prediction. Old records, which were considered ment, and computer applications. In addition, usage of AI
as unbreakable, are reached even by amateurs during their is no exception either. Most of the research concerning AI
training units. These have been enabled especially through in sports domains is focused on the prediction of sports
better nutrition and improved training methods (Bulchan- outcomes, game-time analysis to enable efficient decisions
dani 2012). In all sports, the key to reach high level perfor- during competitions, sports biomechanics for performance
mance lies in the athletes preparation in trainings. There can- analysis, which is facilitated by expert systems and neural
not be a success without proper training planning and cor- networks to gait analysis (Lapham and Bartlett 1995). An-
responding training execution on the highest level (Meżyk other commonly used AI technique is pattern recognition,
and Unold 2011). Nevertheless, training planning is a well- which is for example used in football game analysis and
known problem in the sports domain and only top level weight training (Novatchkov and Baca 2013).
coaches are able to deliver training plans of sufficient qual- However, no literature on sports training planning seems
ity to enable athletes to perform on their very best. Training to be available. As a result, the intention of this article is
planning is a complex process, which is affected by a mul- to outline an ongoing project that deals with utilising auto-
titude of factors that also vary according to different sports. mated planning in sports domains. For that reason there is a
These factors include variables such as athletes predisposi- need to identify what is meant by a sports domain and what
tions, athletes health conditions, competition goals, weather elements will be used as an input for automated planning.
conditions, and others (Smith 2003). The vast amount of At first, there is a need to address what sports explicitly
variables makes planning a process, which depends on ba- means. Collinsdictionary.com (2016) defines sport as:
sic principles that are provided in the coaching literature.
an individual or group activity pursued for exercise or
The purpose of this research is a development of an auto-
pleasure, often involving the testing of physical capa-
mated planning system that uses a specific sports knowledge
bilities and taking the form of a competitive game such
base and desired goals as an input for a planner that will cre-
as football, tennis, etc.
ate individual training plans for athletes. This will provide
efficient solutions for coaches from various sports to create The term sport has many definitions, but all of them sug-
plans without having a vast background knowledge or the gest that sports are primarily concerned with reaching the
need of being an expert in automated planning. In order to best possible performance in a particular discipline. Essen-
reduce the complexity of having very varied requirements tially, athletes in order to achieve required outcomes and
in diverse sports, one specific sports domain, namely kick- performance during competitions, have a need to adequately
boxing, has been chosen as a case study. The chosen domain train before the competition takes place. Athletes are typi-
will serve as a stepping stone and proof of concept for fur- cally guided by coaches, who provide them with knowledge
ther development. in the specific sports domain. This knowledge is usually
transferred by means of training units, which are planned of evaluating and analysing an individual athlete’s execu-
according to coach experiences. In other words, the main tion of a specific task and a level of skill involved in the
objective of sports training is to reach the very best perfor- task. That said, there is a need to evaluate performance in
mance in the planned competition. all sports as it is used for determining competition winners
Sports training can be defined as a set of training meth- and also for sports training improvement. One of the main
ods and exercises, which are executed by an athlete in order purpose of performance evaluation (PE) is to obtain sports
to prepare for a competition. To be able to utilize the best specific data, which are analysed in order to detect errors in
from training, it is necessary to carry out the methods and training process and adopt corrective actions (Owusu 2007).
exercises in precise manners. For that purpose a tool called Higgins (1977) views PE as a complex process that in-
annual training planning has been developed. Nevertheless, cludes numerous stages, which are:
this tool requires vast knowledge and time resources.
• Describing what should happen
Annual Training Planning (ATP) • Describing what happened
Kassa (2011) describes the annual training plan as a tool • Comparing expectations with results
used by coaches, which serves as a base for all scheduled
training activities over a year. The annual training plan is • Taking corrective actions
an important tool for enhancing an athletes performance. Contribution to the view of Higgins’ PE have been done
An ATP is constructed on the concept of periodisation, by Fairs (1987), who described five steps in performance
which divides an athlete’s year into manageable training pe- evaluation:
riods (Milanović and Šalaj 2014). These periods are focused
• Data collection
on development of different abilities such as strength, en-
durance, and speed (Bompa 2009). Furthermore, periodisa- • Diagnostics
tion is a process, which allows to formulate sports program • Prescribed plan of action
in systematic fashion.
Forming an ATP has following steps (Olander, 2015): • Implementation
• Information gathering • Evaluation
• Analysing the last programme The process of PE typically includes collection and analy-
sis of large amount of biased information about athlete’s per-
• Assessing athletes performance
formance. According to Fair’s data collection is fact-finding
• Input the peak events of a year part of PE process, where data are drawn without making
• Dividing year according to periodization any conclusion or interpretation. He claims that the data col-
lected in this step can be of an objective and a subjective
• Outlining objectives of each phase manner. Subjective data are usually provided by an athlete,
• Determining activities of each phase while objective data are collected by evaluator using specific
equipment for explicit measurement.
• Identifying of volume intensity and recovery relationship
The PE procedure is based on several expertises and it
within a season
can differ according to various sports. Further, sports perfor-
• Determine a total number of training hours to be com- mance can be measured in many possible ways and it usually
plete. involves qualitative and quantitative analysis of human mo-
• Identify appropriate training units for each phase tion, coaching methods and biomechanics. The qualitative
analysis of sports performance is based on a visual obser-
To perform all these steps, it is necessary to have a high vation of human motion. This analysis is usually done by
level of knowledge in the field of sports science. Also, it evaluators and accuracy of this method depends on their ex-
is essential to have a great understanding of athletes, who perience and equipment they use. This method is inclinable
are being trained, be aware of all possibilities and circum- to errors as it relies on the evaluator getting a clear picture
stances, which can occur during the season and most of all of joint movements as they occur, which can be in some sit-
possess a great deal of planning abilities. In order to have uation difficult (Fernandes, Anes, and Abrantes 1996). On
accurate and precise input for an ATP and objective out- the other hand, the quantitative approach retrieve an objec-
comes, there is a need to understand athletes’ performance tive data, which have a form of biochemical profile of mo-
and methods by which it can be measured. tion, which is being analysed (Owusu 2007). Nonetheless,
this approach is vastly time consuming. One of the reasons
Performance Evaluation is that biomechanical quantification is a manual process.
One of the best known definitions of sports performance was In addition, Maglischo (2003) describes blood testing pro-
defined by Hay and Reid (1988): performance as the man- cesses for monitoring heart rate with reference to train-
ner in which all movements comprising motor skills – a se- ing managing and monitoring. Also, Friel (2016) describes
ries of voluntary movements of the human body designed to proper and effective methodical approach of making training
achieve a specific goal - are executed. In other words, perfor- logs, which are aimed to create better experience of training
mance is a goal-directed set of movements, and the process preparing phase usage. These two papers are stepping stone
into methodological training and performance testing meth-
ods that are used for the research.
With a rising number of analysed athletes, the complex-
ity can radically increase. This can be seen in situations
when the coach has to remember information about more
than twenty athletes. This information consists of their per-
sonal information (age, weight, height, etc.), number and
performance of athletes competitions, and their strengths
and weaknesses. Retaining and analysing of these informa-
tion brings difficulties, which usually are reflected in train-
ing efficiency, and therefore in performance efficiency.
It is therefore not surprising that numerous computer-
based systems have been developed in order to increase the
speed and quality of performance evaluation (Owusu 2007).
A large number of these systems are visual. The princi-
pal idea of these systems is to capture the complete ath- Figure 1: Proposed system architecture.
lete motion into digital form and then input these visual
data for quantification into computers, where these data are
analysed. Consequently, this provides enormous possibility previous annual training plan. These three parts will be used
for AI techniques. Nevertheless, AI techniques in PE have for generating initial state, objects, and a goal state.
currently only limited capabilities. The main reason is that The next step in the process of planning is fetching the
sports domain lacks of characterisations in terms of quantifi- domain and planning problem into chosen planner and gen-
cation of sports science. This fact is also one of the obstacles erating the annual training plan.
for this research. After the plan is generated, it has to be executed for a
For this reason, there is a need for a sports domain specifi- specific amount of time (chosen by coach). The chosen time
cation that will provide input for a planner, which will result period has to be long enough for the athlete’s body, there-
in generating an annual training plan. In order to create a fore the athlete’s performance, to respond. When this time
sports domain model, there is a need to examine sports per- frame runs out, a performance evaluation process takes a
formance, physical abilities, training methods, training plan- place. Reached performance is compared with the planned
ning and testing of athletes, and adapt this concepts on con- goal state. If the goal state is met, the athlete continues with
cepts of automated planning. plan execution. Otherwise, the reached state becomes the
new initial state and re-planning process is initiated.
Proposed System Architecture
Ongoing Work
The proposed planning system will be developed on an on- The system architecture described above provides a main
tology core that will provide a clear formalism and user- concept of the research idea. Nonetheless, there is still need
friendly environment for domain experts, namely coaches to completely define the sports domain, which would con-
and sports science researchers, to be able to encode sports nect sports training methods, and exercises with their impact
domain knowledge. The encoded knowledge and an input on a human physiology and automated planning concepts.
required for creating initial and goal state will be transferred
into Planning Domain Description Language. The Next Steps
The system architecture (depicted in Figure 1) is based
The first attempt to produce automated training plan will
on two main AI concepts, which are automated planning
be done through a knowledge encoding tool called KEWI.
and ontology. A high-level architecture of the system can
KEWI was developed in order to provide for domain ex-
be seen in figure 1. In proposed scenario ontology outlines
perts, who are not experts in AI planning, a user-friendly
the sports domain itself. In other words, the system ontology
solution for encoding classical planning domains. This tool
will consist of objects definitions, their relationships, and hi-
can export domains into PDDL that can be used with stan-
erarchies, action types and methods. Classes will represent
dard planning engines (Wickler, Chrpa, and Leo McCluskey
individual body elements, such as energy systems, muscular
2014). The concept of KEWI is building on an ontology that
system and technical and tactical capabilities. Action types
furnishes necessary parts of classical planning, such as ob-
will consist of action name, logical preconditions and effect.
jects, their hierarchies, action types, preconditions, effects
Methods will be used as an abstraction of the actions. In
and methods (Wickler, Chrpa, and Leo McCluskey 2014).
other words, methods will be used as high-level tasks that
In order to be able to encode the knowledge into the ontol-
consist of actions which have an explicit ordering.
ogy that is specifically developed for classical planning use,
The input part of the architecture is focused on generating there is a need of major simplification of sports science con-
a planning problem that is going to be created by inputting structs.
three main parts, the first is an athlete’s current performance,
peak events for which the athlete is aiming for, and correc- Example The athlete’s sports performance can be broken
tion values that will be provided by analysis of the athlete’s down into four main categories, which are energy systems,
muscular system, technical and tactical skills. Each activity no sophisticated tool that could provide any level of training
(training method, or simply exercise) has a different impact planning automation. Subsequently, the development of the
on individual abilities of the athlete. The activity impact is sports domain, even though with specific orientation, could
only hardly measurable, which makes it almost impossible provide a tool for a range of sports scientists that would ben-
to define activity with accurate effects. Furthermore, each efit from using this system.
activity has to be executed for a certain amount of time and
be placed in a time frame (training unit). As the domain is References
being developed under classical planning, the time require- Bompa, T. 2009. Periodization : theory and methodology of
ment has to be relaxed. training. Champaign, IL: Human Kinetics.
Bulchandani, S. 2012. Understanding the In uence of Tech-
Challenges nology on Athletic Performance (2). (2):15–17.
The domain has to be concrete in the sports science find-
Fairs, J. 1987. The coaching process: The essence of coach-
ings. The fact that sports science is based on several other
ing. Sports Coach 11(1):17–19.
sciences, such as anatomy, biomechanics, biochemistry and
bio-kinetics, makes the encoding of the domain very chal- Fernandes, O.; Anes, E.; and Abrantes, J. 1996. Biosisttwo
lenging. The domain specification itself brings several prob- dimensional kinematic performance analysis. 14:189–192.
lems that have to be dealt with. These are for example: Friel, J. 2016. The triathlete’s training bible : the world’s
1. Quantification of four different performance abilities. most comprehensive training guide. Boulder, Colorado:
VeloPress.
2. Automation of sports training methods and exercises with
Hay, J. G., and Reid, J. G. 1988. Anatomy, Mechanics, and
correspondence with automated planning concepts.
Human Motion (2nd Edition). Prentice Hall.
3. Addressing activities duration requirement. Higgins, J. R. 1977. Human movement: An integrated ap-
4. Incorporate specific breaks between individual exercises. proach. Mosby.
5. Identify how to prioritize actions. Hou, Z. 2015. The Application of Computer Technology in
6. Identifying of a goal state. Sports Training. Number Etmhs, 209–212. Qinhuangdao:
Atlantis Press.
Currently, different types of automated planning are be- Kassa, M. 2011. Assesment of Annual Training Plan Prepa-
ing addressed, which could satisfy the system requirements. ration in the Case of Addis Ababa Handball Clubs. Ph.D.
Momentarily, the main focus is on classical planning as it Dissertation, ADDIS ABABA UNIVERSITY.
can provide a working prototype of the proposed system.
Nonetheless, the sports domain has some needs, which clas- Lapham, A., and Bartlett, R. 1995. The use of artificial in-
sical planning cannot facilitate. For example, each action in telligence in the analysis of sports performance: A review of
the sports domain requires for an execution certain amount applications in human gait analysis and future directions for
of time. That could be supplied by temporal planning that sports biomechanics. Journal of Sports Sciences 13(3):229–
is allows for expressing time explicitly. Executed actions or 237. PMID: 7563290.
planning operators do not have an immediate effect, but in- Maglischo, E. 2003. Swimming fastest. Champaign, IL:
cur a pre-defined delay. Temporal planning could also pro- Human Kinetics.
vide a solution for incorporating breaks between individual Meżyk, E., and Unold, O. 2011. Machine learning ap-
exercises as each durative action has effects with two sets. proach to model sport training. Computers in Human Be-
One set represents the effects taking place during execution havior 27:1499–1506.
and the other set represents the effects that take a place af- Milanović, D., and Šalaj, S. 2014. Anthropological and
ter the execution of the action. Further, this could be used Methodological (Didactic) Aspects of Work in Contempo-
as representation of athletes’ power supplies. After execu- rary Sports. Sportlogia 10(2):45–54.
tion of an exercise an athlete would require a refreshing of
power supplies in order to execute the next planned exercise. Novatchkov, H., and Baca, A. 2013. Artificial intelligence in
This process would continue until the athlete’s power will be sports on the example of weight training. Journal of Sports
completely depleted or the training session will terminate. Science and Medicine 12(1):27–37.
Owusu, G. 2007. AI and computer-based methods in perfor-
Conclusion and Future Work mance evaluation of sporting feats: An overview. Artificial
In this paper, we have introduced an automated planning ap- Intelligence Review 27(1):57–70.
plication on a real case scenario concerning sports training Smith, D. J. 2003. A Framework for Understanding the
planning. In other words, this paper presents a sports training Training Process Leading to Elite Performance. Sports
planning system architecture and its ongoing development. Medicine 33(15):1103–1126.
This work is carried out in order to develop a planning sys- Wickler, G.; Chrpa, L.; and Leo McCluskey, T. 2014.
tem that will enhance the sports training process. It will cre- KEWI: A Knowledge Engineering Tool for Modelling AI
ate training plans based on the latest sports science research Planning Tasks. 21–24.
and explicit athlete performance data. This would be a sig-
nificant enhancement in sports training as currently there is