Amaury Sports Organization (ASO) and NTT have been powering up their real-time data-monitoring tools during this year’s Tour de France — and adding an AI “digital human” while they’re at it.
The main features being implemented include real-time data monitoring of each rider, spectating crowds, and geographical elements; a digital twin of the entire event connecting the collected data; and Marianna, NTT’s AI- and ChatGPT-enabled digital human solution trained on both real-time and contextual race information for Tour fans.
ASO and NTT have worked together for nearly a decade, and NTT Director Joan Kuhrmann told Silverlinings that while the real-time digital twin feature has been used in years past, this year the company has moved away from a siloed data structure to make lower-latency performance more possible than ever.
NTT’s challenge with this specific “stadium” — one that moves every day of the event — has always been the demands of the network infrastructure across rural France, according to Kuhrmann, but moving into software-defined, edge-to-cloud environments has enabled the necessary performance for effective real-time data monitoring.
“When we look at the timeline between 2015 and 2023, we've really grown that solution from data processing and real-time analytics to build an entire environment with edge computing, cloud, AI, ML capabilities so we can actually connect all of the parts of the race…. This year, we are able to integrate and act on the information when [it was] only available in siloed systems, or not at all, in previous years,” she explained.
Tour de real-time data
According to Kuhrmann, NTT’s service for the Tour de France necessitates “100% uptime” and low latency connection, from the bicycle to the edge connections, then to the cloud where it can be made helpful to both fans and event organizers.
To make that possible, geolocation-enabled sensors are mounted beneath each rider’s saddle transmitting constant latitude, longitude, speed, time gaps and weather data across approximately 3,400 kilometers of France. Microwave signals from the sensors then connect to edge-computing devices on trucks running a containerized version of NTT’s analytics platform. The transmission process happens within milliseconds.
All of the connected real-time data is then used to create a digital twin, replicating “all aspects” of the event, according to the release, including people, operations and geographical aspects to help ASO run the event smoothly.
Tim Wade, CTO of TwinLabs.ai, a sports tech startup launched by NTT last month, explained to Silverlinings that the biggest challenge in “deploying such an extensive range of sensors, diverse infrastructure and services is ensuring we have a real-time view of them. This is both from the aspect of health and additionally from data flowing through them in further layers up the stack.”
But Wade added that having a software-defined set of services and data structures has allowed for greater flexibility and automation in deploying complex services like these. “Of course, the setup and maintenance of the tools in a system [that] is rapidly developing is something that needs to be closely managed.”
On the fan-facing side, Kuhrmann explained the AI digital human tool, Marianne, integrates a combination of hard-coded data, such as general information on competitive bike racing, real-time data collected from the race, and the integration of ChatGPT — effectively enabling her to talk with specifics on the race at any given time.
“It's not for everybody — other people prefer race center, for example — but, if you prefer having [a] conversation, then Marianne is for that particular group of fans,” she noted. “It is basically to give fans another alternative to access race information.”
The men's Tour de France 2023 finishes on July 23, 2023, in Paris, France. The women's Tour kicks off the same day in Clermont-Ferrand, France.