Tackling Tradition: The Integration of Artificial Intelligence in Rugby
Recently, Google Deepmind announced TacticAI, an AI assistant for soccer tactics developed in collaboration with Liverpool FC. It’s not surprising that such state-of-the-art applications of machine learning take place in the world’s most popular sport. However, the rise of artificial intelligence in sports is by no means exclusive to the world of soccer, it is rapidly growing market throughout sports according to Allied Market Research. I have been spending a lot of time in the last decade on a similar field to the Liverpool FC players, but playing a very different sport, rugby. This announcement got me curious about the current machine learning state of affairs in this sport. How is artificial intelligence shaping the present and future way of playing and experiencing Rugby?
Rugby
Rugby union is characterized by its chaotic gameplay, high injury rates, and complex rules that can make the sport hard to follow and predict. Machine learning applications can offer fascinating opportunities to unravel this complexity. This blend of rigorous physicality and advanced analytics is what makes modern rugby union not only a test of physical endurance but also a compelling field for technological innovation in sports. By leveraging machine learning, we can enhance the game for players, coaches, referees, and fans!
Data collection
Before we delve into the applications of machine learning in Rugby, we need to address another challenge first, data collection. Machine Learning algorithms are fully dependent on the data we can provide them with. Therefore, collecting the right data is a crucial step on the way to getting championship winning insights. And it is certainly not as straightforward on a rugby pitch as it is on e.g. the stock market. Let’s look at some ways that have been applied to quantify this organised chaos happening on the pitch.
Wearables
Teams employ wearable technology to monitor players’ physical and physiological data in real-time. Small devices equipped with sensors like gyroscopes, magnetometers, accelerometers, GPS trackers measure speed, distance covered, heart rate, body temperature, and much more. These devices are often embedded into the players’ jerseys or shoes such that they can we worn and collecting data throughout all practice sessions and games. Check out this video to find out more about these wearables.
Video
Recording players has been a popular way to record and review athletes performance during practice for many years. It can provide the athlete with a perspective that they could never get while being in the action themselves. Artificial Intelligence can bring video analysis to a higher level by taking a lot of the manual effort out of the equation. Advanced video analytics platforms apply machine learning on game footage data to track player movements, ball possession, team formations and more. This technology enables coaches to dissect opponents’ strategies and enhance their team’s performance.
Drones provide unique perspectives of the game, capturing data unattainable from ground level. By analyzing drone footage, teams can assess formations and spatial dynamics, offering insights into effective tactics against different opponents. Clubs like Saracens in the English Premiership Rugby have utilized drone technology for a.o. tactical analysis and player positioning improvement. Check out the video below to find out more.
Applications
Now that we have looked into the main data collection methods, it’s time dive into the applications. The selected applications described here are not covering 100% of all the efforts being done by all the rugby clubs and the union within machine learning. However, it should give you a good impression of the variety and range of applications in the rugby world.
Player and team performance
The application of data and machine learning to improve player and team performance is most closely linked to the TacticAI application developed by Google mentioned in the introduction. These application do also exist within Rugby, albeit not at the same level yet. Data from wearables, drones, and video analytics contribute to a holistic view of player and team performance. Machine learning models can predict fatigue levels, optimal play styles, and even susceptibility to specific injuries.
Injury prevention
Injury prevention is a big topic in a physical and sometimes brutal sport like rugby. Therefore, it has also been a domain where researchers have focused a lot of their effort, like this study in Australia that found that their results “suggest that robust machine-learning techniques can appropriately monitor injury risk in professional team-sport athletes”.
A specific type of injury has gotten a lot of attention in recent years, namely head trauma injury. Playing rugby, especially at a professional level, has been linked to a higher risk of brain injury. Not only during their career, but also during the year that follow. Therefore the World Rugby association has invested in the development of smart mouthguard technology. Through sensors they will be able to measure the impact on player’ heads much more accurately, and thereby accurately predict whether or not a player has suffered a concussion.
Scouting
The idea of data-driven player scouting has been popularised with the mass audiences by Hollywood with movies like Moneyball. This is actually the reality nowadays for certain clubs. Algorithms can analyze data from various leagues to spot emerging talents based on specific skill sets, such as tackle success rate or sprint speed. The London Irish, a british rugby Championship club, hired the London based ASI Data Science to revolutionalize the way they find talented players. According to an article in the Sports business journal, the software they developed allows them to find player that have similar characteristics to the ideal player they would want to acquire. For example, they could fill in the name “Jonah Lomu”, arguable the best rugby player of all time, and find players that have similar charactertics that may still be undiscovered gems.
Score prediction
ML models utilize historical data and real-time match statistics to predict scores, enhancing betting markets and fan engagement. An example is the statistical models developed for the Rugby World Cup, which offered fans insights into likely match outcomes, creating a more interactive and engaging viewing experience.
Kick predictor
A collaboration between Guinness Six Nations Rugby, Stats Perform, and AWS uses Amazon SageMaker to deliver real-time, ML-powered insights during rugby matches. The system predicts the likelihood of successful penalty kicks with machine learning models trained and deployed via SageMaker. These insights, which include success probabilities and match stats, are then broadcast live, enhancing viewer experience and understanding of the game’s dynamics.
The kick predictor employs features based on the location on the pitch, player performance, and on the in-game situational such as minute of the match and game score. The location-based features are the most influential of these three. For a more in depth review of the kick predictor, read this blog by AWS.
It’s clear that the applications of Machine Learning in Rugby are not yet at the level of Google’s Tactic AI, but I believe it has proven it’s potential in the rugby world and will continue to have a bigger influence. As the sport continues to embrace these technological advancements, the future of rugby looks both exciting and innovative. I am looking forward to the next development!
