The Power of AI
Over the past few decades, artificial intelligence (AI) has played an increasingly significant role in our daily lives, and this trend will continue. Our mailboxes use it to prevent spam and improve productivity by filtering out unwanted emails. Many search engines use AI algorithms to recommend content that is relevant to our interests and offer us tailor-made plans based on our needs. We can monitor and track our health condition through wearable technology during workouts, helping us exercise to the fullest extent possible. As in Figure 1, AI can even generate images or videos based on a few keywords!
Why AI in Sports?
In the case of traditional sports coaching, we may be able to notice certain obvious downsides. There may be a lack of consistency in athlete evaluation, not to mention the fact that the statistics obtained are often not comprehensive. This is where the application of AI could make a difference. In contrast to the traditional way of coaching sports, we can identify certain patterns that often remain unnoticed yet are crucial to boosting athletic performance. With the help of various kinds of sensors and high-performance cameras, we could collect data generated during workouts, such as the athlete's average heart rate, calories burned, cardio fitness, and so on. Then through AI computation, we can analyze the fitness state of the athlete and make corresponding trainingplans for better results. For example, we can collect the movement data through a wearable IMU sensor and use AI models to recognize the activity type (Figure 2).
Figure 1. Joint detection results by MediaPipe Pose on an AI-generated image with keywords "AI, machine learning, sports, athlete" by Hypotenuse AI.
Figure 2. The signals collected by a wearable IMU sensor.
Research Collaboration with Hong Kong Sports Institute (HKSI)
In comparison with ordinary sports enthusiasts, elite athletes have higher demands on training levels and require specialized and customised training programs. This request has prompted a collaboration project of HK$ 6.7M with the Hong Kong Sports Institute (HKSI). The project team comprises a group of researchers from EdUHK, including Dr. Fu Hong (MIT), Professor Philip Yu (MIT), Dr. Alpha Ling (MIT), Dr. Song Yanjie (MIT), Professor Daniel Chow (HPE), Dr. Sun Fenghua (HPE), Dr. Tong Xiuhong (PS), and experts from other institutes. The cooperative project with HKSI will serve as a model of application-oriented training that can help athletes to improve their training effectiveness and achieve better results in the field of play. Intelligent wearables and sensors are the key project highlights. An in-depth analysis of the athlete's performance can be established based on the data collected during daily training. For example, cameras and motion sensors may be able to locate joint landmarks and be utilized to construct a structural model. By combining it with AI predictive model, we can even calculate the athlete's movement in the next instance. As for our audience -- athletes and coaches, we cannot just give a dazzling string of data. The statistics must be fully visualized, converted into analytical graphs or charts, and presented in a user-friendly manner. Athletes and coaches need to be able to identify each other's strengths and weaknesses so that they can adjust their posture or approach.
Impact of the research
In cooperating with the partners of the Hong Kong Sports Institute, we have discovered their training passion and dedication to sports, as well as their strong spirit of sportsmanship, which likewise drives us to develop automated, efficient and fine granular AI tools for increased effectiveness of their training with coaching. Such interdisciplinary research provides tailored made support to athletes and coaches while also making the sport more accessible to a broader audience. It is anticipated that the AI and the internet of things involved in this research will generate new research directions and impact society.
