Australian swimming athletes achieved three times more gold medals in the 2020 Tokyo Olympics than in the 2016 Olympics with AI technology assistance.
Will advanced artificial technology replace the role of coaches in the future?
YES
OR
NO
The two experts, the designer of the AI system Zhen He, and the Olympics campaign lead for Swimming Australia Jess Corones, reckon that artificial intelligence is only a powerful auxiliary tool for coaches at this stage and in the future, and it cannot completely replicate the coach’s comprehensive capabilities.
With AI technologies thriving around the world, using AI technologies to collect and analyse data is applied to numerous domains for better achievements. Swimming Australia, the peak governing body, applies advanced technologies to elite athlete training and competitions as well. Compared with entirely manual recording data, intelligent technologies can capture swimming athletes’ movements more accurately and faster through cameras, which also challenges parts of coaches’ duties. Therefore, as Sam Isaacson mentioned in his article, whether the role of the coach would be replaced by AI technologies or not is our concern.
Jess Corones has been at the forefront of incorporating AI technology—Sparta 2—into coaching. “The use of Sparta 2 changes the rules of the game,” Corones proposed. “Artificial intelligence improves the decision-making ability of coaches, making it particularly important in tailoring training plans for athletes.”
Since 2015, Swim Performance and Race Tactical Analysis (Sparta 2) developer team has collaborated with the Australian Institute of Sports to establish a competition analysis software that enhances the efficiency of gathering data.
The design of Sparta 2 is inspired by the previous Sparta 1 system which is a traditional analysis method that relies on manual annotation (press the spacebar to pause the video) to track Athletes' performance and can only track one athlete each camera.
“This is not a very efficient use for sports analysts because they should be doing something more useful than just pressing the spacebar.” Zhen He, one of the developers of Sparta 2, mentioned in an interview. In order to deal with this problem, they chose to adopt AI technology to automate the entire analysis process.
The Sparta 2 system is a significant improvement over the previous version. The new system uses a unified 4K camera to capture the movements of all contestants, and uses deep learning algorithms to track and analyse each swimmer's speed, swimming frequency and other performance. This method not only significantly improves the speed and accuracy of data processing, but also greatly simplifies the data collection process.
Unlike competition settings, training pools typically contain up to 40 swimmers, making data tracking complex. By using a single 4K camera to handle all the swimmers' data, Sparta 2 significantly reduces the complexity of data analysis. Analysing swimming speed and stroke frequency helps in designing specific training programs for optimal performance outcomes.
(Chart made by group, data from Wikipedia.)
“Analysing swimming speed and stroke rate helps us tailor specific training sessions designed so that the athletes can get the optimal performance outcome”.
“This level of customisation is not feasible without the assistance of artificial intelligence. Sparta 2 also allows us to gain in-depth insights into our opponents' strategies, tactics, and game data to gain an international competitive advantage”.
Following the success at the Tokyo Olympics, Jess Corones spoke about the further collaboration with AWS, focusing on improvements in data processing and its accessibility, which had a significant impact on performance metrics. Recognising the potential benefits, they have increased their investment in AI technology, aiming to leverage more comprehensive data insights to further improve athlete performance.
“The newly released Lane 4 is like a reactor portal where all the data is stored. It acts as a portal for coaches, allowing them to access all of this data and forecasting through the AWS QuickSight dashboard.”She emphasised that this integration allows for real-time, easy access to essential information from strength to nutrition, making training guidance more efficient during training sessions.
AI technologies excel in data analysis, can self-learn quickly, and provide insightful insights, but they cannot completely replace the role of coaches.
“We're working with humans in swimming. The data can't tell us if the athlete is nervous or has been sick or injured, coaches have to make training adjustments every day based on athlete feelings and status.”
“I think training is psychological. Coaches have to motivate the swimmer to train harder for all of these kinds of things. I don't think AI will do a good job at this.”
Although Sparta 2 has achieved significant breakthroughs at the technical level, it also faced several challenges in the initial stage, especially suggestions and feedback from coaches.
“Some coaches prefer to use traditional teaching methods because they just want to see very clearly what the swimmer was doing.” Zhen He pointed out that we need to find a balance between launching an efficient new system and meeting the user's usage habits.
This interactive visualisation is made by Shorthand
Built with Shorthand