Sports and AI

This year, Japan hosted the Rugby World Cup, and next year, the 2020 Tokyo Olympics will finally be held. In these large-scale sporting events, information and communication technology (ICT) and artificial intelligence (AI) are beginning to be widely utilized. In particular, due to the growing demand for accuracy and fairness in officiating, systems that use sensors and camera footage to assist with calls are now in practical use. Many of you are likely familiar with systems like the ball landing position determination system in tennis and the goal-line technology in soccer. These systems achieve precise officiating by measuring the ball's position through image analysis from multiple camera angles. It can be said that they provide a new set of "eyes" to assist with officiating.

In our laboratory, we are jointly developing a system with Toshiba that uses image processing and deep learning to automatically classify major plays necessary for tactical analysis in rugby, using footage from a single camera (Figure 1). Using deep learning, the system detects and tracks players and the ball in the footage, maps the camera's view to the ground, and obtains the coordinates of the players and the ball on the field. Then, using their positional relationships and movements as features, it automatically classifies plays such as passes, scrums, and kicks. This allows team analysts to use the automatically classified statistical information after a match to focus on more advanced tactical analysis.

Let's consider where this research might ultimately lead. For officiating assistance systems, this could mean the realization of "robot referees" that make calls and score based on objective observation and data analysis, but for now, their role remains strictly to "assist" in officiating. While some sports may already be technically automatable, I believe that both officials and spectators are still somewhat reluctant to fully entrust officiating to machines. The same applies to tactical analysis. In the rugby example mentioned earlier, the current system only goes as far as quickly providing statistical data to analysts; the subsequent detailed tactical analysis and specific strategy formulation remain the job of expert analysts. As data from numerous matches accumulates, it will likely become technically possible to create an AI that can even propose new tactics. The time may come when the team with the smarter AI technology consistently wins. Personally, I believe the ideal is a collaborative relationship where humans use the various analytical data provided by AI to engage in more creative intellectual activities.

The annual rugby Waseda-Keio rivalry is always a close contest and generates great excitement. I go to watch it every year, and I feel that even as times and people change, the traditional rugby playing styles of Keio and Waseda are continuously passed down. While data is effectively utilized as data, there is no doubt that something unrelated to AI is at work, something that stirs the emotions of the players on the field.

In any case, it is not uncommon for technological innovations to arise from major events like the Olympics and the World Cup. Why not enjoy watching the upcoming Tokyo Olympics from a slightly different perspective, paying attention to what new technologies will be utilized?