Abstract, Distorted Computer Motherboard_edited.jpg

Winter Olympics Use Cases: New applications for Artificial Intelligence

Since the 2022 Winter Games are live in Beijing, we thought it might be fun to look into this space for inspiration. What kind of inspiration? Clever use cases for AI product solutions can be found in any industry vertical. The innovation can come from seeing the use in one scenario and applying it to another.


At this year's Olympic Games, we found examples of AI being used to monitor extremely detailed weather conditions, assist with training and coaching, analyze scoring accuracies and of course, connect with the fans around the world.


> Jump ahead to see the 4 AI use cases from the Beijing Winter Olympics.


How data science strategies connect product ideas


In IT product development, it’s pretty common to see a great business model or application of some tech and think “What if...”.


Gravitate helps customers find those “what if” opportunities and looks for ways to apply novel data science strategies to challenging business cases and product road maps.


And what is more compelling testing ground for new ideas right now than the Winter Olympics? It is considered the ultimate playground for experimenting with the latest advancements in sports science -- from harnessing data to improve an athlete’s performance, to finding new ways to connect with spectators on-site and virtually around the world.


In fact, research from Deloitte anticipates the rapidly growing sports analytics industry to reach nearly $4 billion by 2023. So it’s a good time to learn from this disruptive, quickly evolving space.

Gravitate recently partnered with sports analytics company Realplay Sports to help players improve their skills. We decided to bring some of our new interests into the spotlight through this round-up of AI examples inspired from the 2022 Olympic Games and Paralympic Games.


AI product inspiration from 4 Olympic use cases


We believe the data science strategies used at this year's Olympic Games can inspire SaaS startups to augment customer experiences and business intelligence. Product development teams can enjoy technical explorations of algorithms, while others looking to understand AI use cases can glean new perspectives for your own startup needs.


Monitoring precise weather conditions with AI


Outdoor sports are completely vulnerable to weather conditions. For several Winter Olympic sports, those details become even more critical. For example, a research team focused on the most common snow sports that share specific requirements for wind speed and direction, with some having to watch for maximum wind gusts or altering directions. For the athletes in these sports, monitoring the weather every 10 minutes is a logical priority problem to solve.


What caught our attention?

The output machine learning (MOML) post-processing method increased the forecasting accuracy by more than 10%.

Solving gaps in time with data science

It’s important to note that many of these locations were built just for the Olympic Games. This means there is little history of weather data to incorporate into predictive models. And that is where the extra layers of machine learning come into play. Researchers were interested in “extending the forecasting system’s ability to automatically learn and improve from past experience."


Read: Machine Learning−based Weather Support for the 2022 Winter Olympics


3D + AI takes athletes beyond one-dimensional analysis


In the Tokyo summer games, Intel led a project called 3DAT using 3D pose estimation to follow the track and field athletes through their movements. Those studies opened the door for considering new ways to improve accuracy both for training athletes and for judging the competitors’ form.


A somewhat controversial example is how AI can provide assistance to referee scoring using 3D + AI technology.


Through quantitative analysis of continual live video for the skiing events, technology from a Chinese-based company Baidu is calculating precise measurements, overlaying data from an athletes’ rotation angle, jump speed and height to render a data-driven assessment of the athlete’s performance.


What caught our attention?

They have formed a skeleton by annotating 22 key locations on the body pixel-by-pixel for automatic tracking. The number of features measured, including velocity, body angle and strike length, are not characteristics the naked eye can easily identify.


AI working together with sport science

Securing a 360 degree view of the skiers’ movements is allowing for side by side comparison. There are infinite uses for this concept of assessing technical differences in detail in almost real-time.


Read: Chinese Tech Giants Serve Beijing Winter Olympics with '3D + AI' technology


Machine learning uses for athlete training and strategic planning

For the past few years, scientists and coaches have combined forces to put machines and deep learning to work identifying opportunities for athletes to improve their form or avoid injury.

One example from 2020 applied multivariate statistics to the Ladies Mass Start speed skating training, and in 2022, the Chinese Olympic curling team trained with an ice-savvy robot.


Even now, coaches and athletes are equal parts curious and skeptical of the power that predictive and even prescriptive analytics can offer for overcoming their training challenges.


What caught our attention?

Paralympics athletes had success in 2020 with AI-based training plans, and yet there isn’t more talk about it now in 2022. Maybe their training secrets will be revealed after competition ends later this month.


What’s the take-away?

Many coaches and athletes are interested to have the insights, but some aren’t convinced data science can replace their experiences. A former Olympian aptly described that unique balance in this interview. It's a worthwhile read that offers good positioning for SaaS business owners faced with a resistant audience fearful of losing jobs to machines.


Read: Sportlogiq’s Craig Buntin on his Olympics journey to analytics, giving coaches data and why numbers can’t tell you everything about an athlete


Connecting with fans, virtual and in-real life


Potentially the hardest challenge for this year's Winter Olympics, was not being able to have spectators attend in person.


Data science has been the hero in overcoming distance. Olympic viewers around the world are able to experience ski jumps in a fully-immersive virtual reality. Through 3D reconstruction technology we can experience all major competition venues.


For those at the Winter Olympic Village needing real-time help navigating their location, this year’s destination has been fully mapped in AR with a browser-based smart augmented reality navigation (AR) application.


What caught our attention?

Data science is also aiding those with hearing disabilities who are attending the Beijing Games through the first AI sign language presenter. The AI robot combines a natural action engine and a sign language translation engine to provide real-time accurate translations.


What to look for going forward?

Identifying patterns is one of the core tenants of machine learning. So it’s a natural fit for Olympic operations to explore how to increase efficiencies in areas like digital ticketing and managing crowds.


According to a Deloitte whitepaper, future stadiums built for Olympic-use will be “technological fortresses and data templates.”


Also watch for more uses of computer vision technology and finely-tuned scoring algorithms, including robots becoming Olympic judges, and the related controversies currently under debate.


Get inspired -- how AI impacts the Olympic Games


With more than 3 Billion people tuning into the previous Olympics on TV or digitally, it’s safe to say, whatever projects and trends are happening in AI, there is something for everyone to learn from at one of the world’s largest events.


Many SaaS product ideas start with a concept that depends on machine learning algorithms. Some startups know they need and want artificial intelligence to be part of the solution, but figuring out where to optimize the results can take time and a deep understanding of data science best practices.


The Gravitate development model augments product and technology teams with a mix of industry expertise and customized skills to scale at the right pace.

As a startup's chosen development partner, we plan and build AI solutions side-by-side. Our vision is to democratize the use of data science and machine learning wherever possible.


We'd love to explore AI product solutions with you for whatever problem your startup is looking to solve.



Read about some of our recent client projects or learn about our agile data science methodology.