From 3M Health Information Systems
AI Talk: Wimbledon analytics and powering Internet of Things
Fourth of July weekend is a time for celebration and relaxation. It is also the midway point for a 140-year-old tradition: Wimbledon. This two-week extravaganza, replete with countless matches in grass, is also a hub for the very best in tech. IBM has partnered with Wimbledon for a number of years (decades apparently) and each year they have added new features to the analytics platform.
A few years back they introduced automatically curated highlights. These video clips, which require a bit of manual proofing, are automatically collected based on noise level, player emotion (a fist pump, a roar) and crowd cheering. Lots of highlight videos and shots of the day are available on the Wimbledon website and the smartphone app.
New for this year? An AI-based algorithmic ranking of each player and pre-match insights predicting which player is likely to win given how they are currently playing. You can look at these predictions and see how many are correct on a match-by-match basis. I wonder what the overall success rate is with these predictions? I noticed that the AI picked American teenager Coco Gauf to win against the three-time grand slam champion Angelique Kerber. Coco lost in straight sets, but the algorithms seem to get it right more often than not.
There is also a new statistic regarding momentum shifts using visual cues to determine how the momentum is shifting with each game and each point.
Another new offering for this year is free Wi-Fi – with a limited rollout. Hotspots are marked by the traditional Wi-Fi symbol that you see in smartphones with attendant signal strength. The data collected from this includes who is using what app and where they are physically located – not for any creepy use, but to plan out where to put a bathroom or strawberry stalls and how to avoid congestion.
As the players navigate the court, slip-sliding every which way, the technology is steadily getting better. Perhaps they need to develop a fall risk solution. The players would love that.
Powering the Internet of Things (IoT)
Over the holiday weekend I was skimming through Communications of the ACM and saw this news item regarding the future of the Internet of Things (IoT). People like Jensen Huang, CEO of NVIDIA, predicts that we will see trillions of AI-powered IoT devices. That is a “trillion” with a “T.” What are these devices? Everything you see around you from toasters to keys to light switches to any object you own which can be smart. That might mean trillions of batteries eventually clogging up your landfill. Not a pleasant thought, nor eco-friendly. So, a set of brave researchers have embarked on a quest to design battery-free IoT devices.
Josiah Hester and researchers at Northwestern University have designed a battery-free Nintendo Game Boy powered by sunlight and button presses! The mechanism harvests energy from the movement of tiny magnets through wound coils whenever someone presses a button. This team is also developing a face mask that harvests energy from a person’s breathing!
Carnegie Mellon University researcher, professor Brandon Lucia and colleagues, have developed a small, battery-free camera that works by harvesting energy from radio waves. They are also developing a software platform that can work with intermittent power – which is going to be the reality for devices that rely on harvesting energy. What is a sample application this device can enable? Cameras distributed throughout cities to monitor traffic and identify congestion hotspots.
Professor Shyam Gollakota of University of Washington and his team are developing solutions that rely on what is called “ambient backscatter.” This is a technology that enables communication between devices by leveraging ambient signals present near the device. The team is working on a battery-free phone. As I spend hours charging my latest iPhone, the prospect of having a phone that I don’t have to charge seems like hallucination. They do have a cool video in the Communications of the ACM article which demos how the concept actually works. It has a lot of potential useful applications involving powering low-powered IoT devices; but powering an iPhone or Android device is probably not feasible, at least in the near future.
MIT Technology Review has just released a new report on “35 innovators under 35” and one of the inventions was created by Dr. Amay Bandodkar of North Carolina State University. Dr. Bandodkar has developed a self-powered biochemical wearable sensor that is powered by a lightweight battery that runs on sweat!
It is quite amazing to see the range of research efforts on how to power IoT devices without traditional batteries.
I am always looking for feedback and if you would like me to cover a story, please let me know! Leave me a comment below or ask a question on my blogger profile page.
V. “Juggy” Jagannathan, PhD, is Director of Research for 3M M*Modal and is an AI Evangelist with four decades of experience in AI and Computer Science research.