Traditional methods of sports coaching have many strengths, but there's one thing they all struggle to do well: provide real-time feedback. The default of coaching is to give detailed feedback only after practice sessions, which makes it difficult for athletes to attempt immediate adjustments to their movements. Having a delay between the initial action, followed by coaching, then trying it a different way, can lead to inefficient training and less-than-optimal performance. But as sports coaching becomes more and more driven by data and statistics, there's a growing demand for technologies that offer both immediacy and precision in how they guide athletes.

That's where NTT's Embodied Knowledge AI comes in. It's a new way of thinking about sports training that bridges the gap between delayed feedback and real-time coaching.

Immediate, Accurate Analysis and Feedback

Developed using a combination of large language models (LLMs) and visual language models (VLMs), and demonstrated at the NTT R&D Forum event recently held in Tokyo, NTT's new tech is designed to offer effective, intuitive feedback during training sessions. Unlike traditional coaching tools, which mostly focus on giving verbal feedback after the fact, Embodied Knowledge AI uses both verbal and non-verbal cues to provide real-time guidance. It's accurate, it's immediate, and it has the potential to transform how athletes learn and refine their techniques. It can give corrections when they matter most—during the activity itself.

The Right Movements, Always

It works through a proprietary movement analysis capability. The AI is able to capture and extract critical frames from an athlete's motion, then compare them to ideal movements, working out where discrepancies are taking place and generating detailed explanations. The process is enhanced by VLMs, which specialize in interpreting visual data, ensuring that the feedback is precise and actionable. The system also uses LLMs to generate multimodal coaching, combining spoken instructions with audiovisual and tactile feedback. It's a comprehensive package that makes the coaching process not only accurate, but also fun and easy to understand.

Bad Habits? Computer Says No

So how would it work for athletes? A swimmer practicing in a pool could get immediate corrections on their stroke technique, lessening the chance of inefficient movements becoming ingrained habits. In the same way, a gym user could adjust their posture or movement during a workout based on tactile prompts or audio cues from the AI system. Because errors are noticed and addressed as they occur, the technology helps ensure faster improvement and a deeper understanding of proper techniques.

Better, More Accessible Coaching

The sports industry should be prepared for what Embodied Knowledge AI can bring, not least of which is reduced coaching costs and better accessibility. Personal trainers may not always be available or affordable. AI systems could deliver expert guidance, 24-7. NTT projects a 30% reduction in human coaching costs, making high-quality training more accessible to amateur athletes and fitness enthusiasts. Indoor gym trainers are planned to be available in 2026.

... and Better Health

The healthcare sector will also benefit. For older adults, maintaining physical activity is crucial for health and independence; Embodied Knowledge AI could deliver personal trainers tailored to the needs of the elderly, offering guidance for exercises that improve balance, strength, and overall well-being. NTT is planning to roll out applications of the technology to include elderly trainers and outdoor activity monitors by 2028.

From guiding a slow jogger through optimal pacing to helping a professional athlete perfect their form, Embodied Knowledge AI's potential is huge. And it's more than just a technical breakthrough; it's a shift in how we approach learning and skill development in physical activities. Better coaching that's personalized and accessible to all.

For more details of the NTT R&D Forum event, please see this website:
https://www.rd.ntt/e/forum/2024/
For further information on NTT's embodied AI for sports coaching technology, please see:
https://www.rd.ntt/forum/2024/doc/D01-14-e.pdf
If you would like to contact a member of NTT's R&D team, please see:
https://tools.group.ntt/en/rd/contact/index.php

NTT—Innovating the Future

Daniel O'Connor

Daniel O'Connor joined the NTT Group in 1999 when he began work as the Public Relations Manager of NTT Europe. While in London, he liaised with the local press, created the company's intranet site, wrote technical copy for industry magazines and managed exhibition stands from initial design to finished displays. 

Later seconded to the headquarters of NTT Communications in Tokyo, he contributed to the company's first-ever winning of global telecoms awards and the digitalisation of internal company information exchange.

Since 2015 Daniel has created content for the Group's Global Leadership Institute, the One NTT Network and is currently working with NTT R&D teams to grow public understanding of the cutting-edge research undertaken by the NTT Group.