MIT News
Q: Why apply AI to figure skating?
Lu: Skaters can at all times keep pushing, higher, faster, stronger. OOFSkate is all about helping skaters determine a strategy to rotate a little bit bit faster of their jumps or jump a little bit bit higher. The system helps skaters catch things that perhaps could pass a watch test, but which may allow them to focus on some high-value areas of opportunity. The artistic side of skating is far harder to guage than the technical elements since it’s subjective.
To make use of mobile training app, you simply have to take a video of an athlete’s jump, and it’s going to spit out the physical metrics that drive what number of rotations you’ll be able to do. It tracks those metrics and builds in all the other current elite and former elite athletes. You possibly can see your data after which see, “That is how an Olympic champion did this element, perhaps I should try that.” You get the comparison and the automated classifier, which shows you when you did this trick at World Championships and it were judged by a global panel, that is roughly the grade of execution rating they might provide you with.
Hosoi: There are a variety of AI tools which might be coming online, especially things like pose estimators, where you’ll be able to approximate skeletal configurations from video. The challenge with these pose estimators is that when you only have one camera angle, they do thoroughly within the plane of the camera, but they do very poorly with depth. For instance, when you’re attempting to critique any individual’s form in fencing, they usually’re moving toward the camera, you get very bad data. But with figure skating, Jerry has found considered one of the few areas where depth challenges don’t really matter. In figure skating, you might want to understand: How high did this person jump, how again and again did they go around, and the way well did they land? None of those depend on depth. He’s found an application that pose estimators do rather well, and that doesn’t pay a penalty for the things they do badly.
Q: Could you ever see a world wherein AI is used to guage the artistic side of figure skating?
Hosoi: In the case of AI and aesthetic evaluation, we now have recent work underway due to a MIT Human Insight Collaborative (MITHIC) grant. This work is in collaboration with Professor Arthur Bahr and IDSS graduate student Eric Liu. While you ask an AI platform for an aesthetic evaluation reminiscent of “What do you think that of this painting?” it’s going to respond with something that seems like it got here from a human. What we would like to know is, to get to that assessment, are the AIs going through the identical kind of reasoning pathways or using the identical intuitive concepts that humans undergo to reach at, “I like that painting,” or “I don’t like that painting”? Or are they simply parrots? Are they simply mimicking what they heard an individual say? Or is there some concept map of aesthetic appeal? Figure skating is an ideal place to search for this map because skating is aesthetically judged. And there are numbers. You possibly can’t go around a museum and find scores, “This painting is a 35.” But in skating, you’ve got the info.
That brings up one other much more interesting query, which is the difference between novices and experts. It’s known that expert humans and novice humans will react in another way to seeing the identical thing. Anyone who’s an authority judge can have a unique opinion of a skating performance than a member of the overall population. We’re trying to know differences between reactions from experts, novices, and AI. Do these reactions have some common ground in where they’re coming from, or is the AI coming from a unique place than each the expert and the novice?
Lu: Figure skating is interesting because everybody working in the sector of AI is attempting to determine AGI or artificial general intelligence and attempting to construct this extremely sound AI that replicates human beings. Working on applying AI to sports like figure skating helps us understand how humans think and approach judging. This has down-the-line impacts for AI research and corporations which might be developing AI models. By gaining a deeper understanding of how current state-of-the-art AI models work with these sports, and the way you might want to do training and fine-tuning of those models to make them work for specific sports, it helps you understand how AI must advance.
Q: What’s going to you be waiting for within the Milan Cortina Olympics figure skating competitions, now that you just’ve been studying and dealing on this area? Do you think that someone will land a quint?
Lu: For the winter games, I’m working with NBC for the figure skating, ski, and snowboarding competitions to assist them tell a data-driven story for the American people. The goal is to make these sports more relatable. Skating looks slow on television, nevertheless it’s not. The whole lot is purported to look effortless. If it looks hard, you might be probably going to get penalized. Skaters have to learn the right way to spin very fast, jump extremely high, float within the air, and land beautifully on one foot. The info we’re gathering can assist showcase how hard skating actually is, though it’s purported to look easy.
I’m glad we’re working within the Olympics sports realm since the world watches once every 4 years, and it’s traditionally coaching-intensive and talent-driven sports, unlike a sport like baseball, where when you don’t have an elite-level optical tracking system you aren’t maximizing the worth that you just currently have. I’m glad we get to work with these Olympic sports and athletes and make an impact here.
Hosoi: I actually have at all times watched Olympic figure skating competitions, ever since I could activate the TV. They’re at all times incredible. One in every of the things that I’m going to be practicing is identifying the jumps, which may be very hard to do when you’re an amateur “judge.”
I actually have also done some back-of-the-envelope calculations to see if a quint is feasible. I’m now totally convinced it’s possible. We’ll see one in our lifetime, if not relatively soon. Not on this Olympics, but soon. Once I saw we were so close on the quint, I assumed, what about six? Can we do six rotations? Probably not. That’s where we start to return up against the bounds of human physical capability. But five, I feel, is in reach.
