Elite swimmers require thousands of training programs a year to prepare for pinnacle performances at international competitions. Coaches have to create several programs per day-, which is not only time consuming, but it also tests the creativity of the authors. Coming up with programs that work towards a specific training goal and that are still somewhat entertaining is difficult. A first implementation of a training program editor is available at https://bartneck.github.io/SwimDsl/
The goal of this project is to create an automatic training program generator that uses principles from generative design, planning and artificial intelligence. The students will identify parameters that need to be systematically manipulated throughout a training season. The next step is to create an algorithm that creates plans that work within these constraints and that offer variations to keep the program interesting. The programs themselves will use swiML to formally describe them, but their parameters, such as intensity and length, will need to be adapted inline with the overall training goals. For example, the length of the programs might increase towards the end of a season. The system should be able to generate hundreds of training programs that are each scheduled and parameterized.
The creative variations can come from using Perlin Noise. It would allow us to procedurally generate programs that have components that are not just random.
Supervisors
Primary Supervisor: Christoph Bartneck
Does the project come with funding
No - applicants must be self-funded or can apply for relevant scholarships
Final date for receiving applications
Ongoing
How to apply
Please send your CV and academic transcripts to christoph.bartneck@canterbury.ac.nz
Keywords
swimming, training, generative design, computer science, sports science
