Artificial Intelligence (AI) is already working with things that surround us. From pretty straightforward applications, such as algorithms recommending music or films based on users’ preferences, to citizenship management software programmes used by governments. AI is also being used in professional sports, where the biggest challenge is providing these programmes with a database to facilitate learning.
This is why AI is the umbrella term for different techniques having a common feature: being self-learning software programmes. To do so, it is necessary to have access to a large database containing large amounts of data, which humans would not be able to handle. Thus, designing facial recognition systems gets relatively easy when considering the millions of photos shared on social media. On the other hand, finding data about professional sports gets difficult as this number is not higher than that of the UNI World Athletes: 85,000 athletes. On top of it, measures can only be useful if they are based on highly specific aspects.
Like everything else in sports, this is more of a challenge than a problem. The methods that some start-ups are adopting to deal with this seem to reveal the approach of AI in sports: to design devices that provide data for learning and can be applied for large groups such as professional athletes. In addition, others, such as those in military training, might also benefit from them since the situations they go through can resemble those of sports, e.g., tough challenges and individual and group workouts.
A Plate which analyses your movement to see if it can Lead to an Injury
One of the most remarkable cases is that of Sparta Science. Phil Wagner, its founder, was an American football player in college who got his athlete career cut short by injuries. After graduating as a physician, he focused to assess the risk of musculoskeletal injury, which is the direct result of repeated minor trauma. For a long time, medical experts have known there is a measurable correlation between the way one moves and the injury risk. Therefore, AI processing data could be the means to identify the risks of kicking, jumping, tackling, or maintaining a poor posture. Precisely, the most meaningful aspect of this case is the device designed to fulfil such objective: the “force plate.”
Using force plate for collecting data at Sparta Science facilities.
When the user jumps, pushes or swings on the plate, it takes just 20 seconds to record 60,000 points of data about them. Several teams of the American National Football League—a sport with high incidence of injuries—have adopted this technology. Thanks to it, their database already includes information about various elite players. NBA and MLB teams also rely on this, but the most promising aspect is the US military, which hired this organization to treat their soldiers’ injuries and save millions of dollars. Through this business relation, there will be a big database available on musculoskeletal injuries and movement in the sports world. After all, the severity of military training and that of professional sports are undoubtedly related.
Objectively Assessing Football Players’ Skills
For the company Seattle Sports Science, an athlete’s ability has never been actually measured. This is defined by the relationship between ball touches by any part of the body and the result obtained after a kick. They identify twenty-three contact point —head (four), trunk (five), legs (two) and feet (six for each foot)— and assess the touch effectiveness. Its purpose is designing football-specific skills analytics.
For such case, Isotechne Motion —a post-shaped embedded sensor device which can get stuck in any training ground— is used for collecting data on impact zone, movement, speed and shooting accuracy. Designed for training, it is combined with a ball throwing machine which launches the ball at different trajectories, speeds and distances. Here, the role of AI is grouping the data collected and the hundreds and thousands of touches made by players for subsequent analysis. AI can self-learn and assess each players’ ability as well as their kick style compared to other peers, passes, kicks to goal, etc.
Here, there is a substantial benefit for both the assessment of new talents and the coach. The latter can precisely know their players’ weaknesses and overcome them by pointing out cases like controlling the trajectory of the ball when striking it with the chest. By analysing an applicant’s performance, it would also be possible to get a range of precision and effectiveness regarding the way teams make use of such information. Moreover, the integration of all this would turn out to be valuable big data for machine learning and AI by identifying useful patterns for all players around the world, or, in other words, by objectively measuring any football player on a scale of 1 to 10 based on their ball touches.
However, this is still a partial glimpse since less skilful players whose tactical knowledge or positioning are better are not considered. Like in many other areas, AI still needs further development and more learning scenarios.
In relation to this, the CEO of Seattle Sport Sciences, Jeff Alger, has a worth-highlighting opinion on AI. According to him, algorithms—the basis of AI—come and go, but data is forever. It could also be pointed out that much of the teams’ and professional players’ success will depend on their ability to have access to such data and process it. Welcome to artificial intelligence in sports.