Master's project: Quantitative biomechanics in trampoline gymnastics and TeamGym trampet
Skills:
UX design, conceptualization, signal processing, programming
Year:
2023
//About the project
The purpose of this project was to understand the fundamental biomechanics involved in skills performed on different trampolines, identify key performance indicators, and investigate relevant performance analysis tools, with the end goal of designing a learning tool for training sessions.
The project consists of four main parts; preliminary research, development of a model for simulation of human aerial motion, acquisition of motion data from skilled athletes, and development of a concept for the learning tool.
The scope of the project was limited to trampoline performances in two gymnastics disciplines; trampoline gymnastics and the trampet discipline of TeamGym.
.01 Research questions & goals
Two primary research questions were considered:
" How can different techniques for the execution of skills in trampoline gymnastics and TeamGym trampet be quantitatively analyzed and compared by coaches and other stakeholders who are not experts in biomechanics? "
" How can a quantitative performance analysis technology be implemented in trampoline gymnastics and TeamGym trampet training sessions in a way so that it provides value to gymnasts as well as coaches while taking into account the learning context and environment? "
In an effort to answer the above research questions, four goals were set:
To identify and provide a fundamental overview of the most important biomechanical indicators of performance in the different phases of a performed skill.
To provide a framework for evaluation of different techniques based on biomechanical modeling and simulation.
To design and develop a concept for a learning tool that captures quantitative performance measures and communicates technical feedback to athletes and coaches.
To demonstrate the feasibility of implementing the developed concept in the real world.
02.Preliminary research
Research was focused on investigating how performance analysis is done in other sports, evaluating relevant motion analysis technologies, understanding aerial motion mechanics and identifying key performance indicators.
Field work was carried out in order to assess training session structure and culture, understand problems and challenges coaches face, and analyze the use of digital technologies during training.
Based on findings from the conducted research, four secondary and more specific research questions were devised:
Can a simulation model based on a simple model of the human body and requiring only basic input variables be used for comparative studies of technique?
Is it possible to make such a simulation framework so easy to use that an average coach is able to?
Can a single inertial sensor provide sufficient data and quality of data to extract biomechanical measures and performance indicators relevant in training?
Can a single inertial sensor be combined with a video capture tool to provide visual feedback as well as biomechanical performance measures in a way that will not require significant effort or time from coaches?
.03 Simulating aerial motion
A simulation framework was created for simulating motion resulting from movement of body limbs while airborne, based on the principle of conservation of angular momentum.
The human body was modeled as 11 cylinders connected by joints with 3 degrees of freedom, to enable simulation of all common body movements. Size, weight and center of mass of each cylinder is modeled relative to specified height and weight of the athlete, based on body scanning data of college aged males and females.
After specifying athlete height and weight, the angular momentum vector acting on the athlete, time of flight, and body positions at specific time points during the skill, along with other parameters, the simulation framework calculates the resulting somersault, tilt and twist angles throughout the skill and can export an animation visualizing the skill.
04.Measuring aerial motion
Talented trampoline and TeamGym athletes were recruited for measuring sessions at a motion capture facility on DTU. Athletes were equipped with a motion tracking suit with attached optical markers and a single IMU sensor with gyroscope, accelerometer and magnetometer. They were then asked to perform a series of specific skills. These skills were recorded on video from two angles (front and side), while the motion capture system recorded marker positions and sensor data was saved.
After the recording sessions were carried out, motion capture data and sensor data was labeled and processed and then synchronized with the video feeds. Then, selected skills were simulated using the developed simulation framework using videos as reference. Finally, the processed data from the motion capture system and sensor, as well as the resulting output from simulations, was compared in order to evaluate the quality of the sensor data and validate the simulation model. The optical motion capture data was used as reference due to the remarkably high precision of the system.
.05 Results and take-aways
After significant processing, the resulting sensor data was deemed sufficiently reliable for determining the performed somersaults and twists and for calculating several relevant performance indicators, such as time of flight, timing of twist initiation and opening, and more. Likewise, the output from the simulation framework was deemed sufficiently accurate for biomechanical analysis of performed skills, including estimation of the angular momentum vector.
Based on the results, along with take-aways from the research phase of the project, the combination of quantitative insights possible with the single sensor and qualitative insights provided by video feedback was found to constitute a strong basis for further design and development of a learning tool concept.
06.Concept development
An ideation workshop was carried out with 4 current and past high-level TeamGym and/or performance gymnastics coaches. The coaches were introduced to the research questions and the problem space, then asked to brainstorm problems and challenges experienced from the perspective of a coach and a gymnast. These challenges were clustered by affinity mapping.
Subsequently, the coaches were introduced to the results and take-aways from the project research and experiments, identified performance indicators, as well as the defined boundaries, scope and requirements. They were asked to brainstorm useful and/or fun measures and features for a potential learning tool solution consisting of a single inertial sensor and a video feedback device, taking into account the identified problems and challenges, their own experience as coaches, and practicality of use during training sessions.
Ideas for measures and features were clustered, combined and further detailed, and three concepts with different focus areas were synthesized.
.07 Prototyping & testing
After the workshop, the most simple concept was chosen for further work and testing. The concept consisted of an iPad app, some sensors and a height adjustable tripod for the iPad.
A simple prototype of the app was created, and the setup was brought to a TeamGym training session for feedback and practicality testing.
Athletes were introduced to the idea, and told that the prototype was not truly functional yet. They were then asked to wear a sensor and go through the process of performing a skill, going to the iPad, and clicking through the prototype app, then to 'talk out loud' about what they were thinking.
The athletes expressed excitement for the solution and provided further ideas for features and functions, including ideas for gamification. In terms of practicality, no issues were found except for a slight discomfort with the sensor attachment mentioned by one gymnast. The iPad was able to record the hurdle jump and performed skill while being placed relatively close to the trampet using the wide-angle camera, and navigating the simple app prototype was intuitive for the majority of gymnasts.
08.Reflection & conclusions
The concept could have been developed much further and in much more detail. However, ensuring proper foundational research of problems and challenges as well as the potentially usable technologies for performance analysis, capturing real-world data from performed skills, and providing a proof-of-concept validating the viability of using a single sensor for quantitative data capture was prioritized over finalizing the learning tool concept.
As such, the data captured and insights gained during this project provides a basis for diverging in the solution space, and many potential future solutions are imaginable.
Furthermore, the data captured can provide further insight into the biomechanical characteristics of specific skills, if furhter analysis is performed. For example, determining the difference in real-world angular momentum produced during take-off for tucked somersault skills (e.g. a triple tucked somersault with half twist) versus straight somersault skills (e.g. a double straight somersault with 1.5 twists) would be useful to know for coaches when teaching athletes how to perform optimal take-offs for the specific skills they are practicing.
