What if you could finally answer: “How many inches make or break a judo throw?” This episode of The Shintaro Higashi Show (guest-hosted by David Kim and Dr. Peter U) unpacks the dream—and the challenges—of blending artificial intelligence with the art of judo.

Judo as a Game of Inches: Science Meets the Mat

Judo is famously called a “game of inches”—but what exactly does that mean? Is it one inch, half an inch, or three? Inspired by shows like Fight Science and the endless “how much force?” questions in sports, David and Dr. Peter ask if there’s a way to actually measure the mechanical secrets behind throws, balance breaking, and the fine details of world-class technique.

Watch the full video here ↓

The Dream: Simulating Judo with Tech

Imagine:

Modeling gravity, body structure, and joint elasticity on a computer.
Capturing real throws in high resolution video or motion capture and then running simulations to analyze force, center of gravity, and optimal technique.
Creating a Platonic “ideal” judoka model… that you can tweak for your own build and style, using AI to predict your best moves!

This would let athletes, coaches, and even commentators quantify everything from kazushi (off-balancing) to grip-fighting, transforming how we learn and optimize judo.

The Roadblocks: Video Data, AI Challenges, and More

Dr. Peter explains why this futuristic scenario is much tougher than it sounds:

AI struggles more with video than text. Videos are information-light and require complex models to learn what matters (and what to ignore) from frame to frame.
Embodied AI and Physics: Unlike language, physical “embodied” AI needs a deep understanding of real-world physics, which is hard to simulate with all the variables of real grappling.
Mocap & Templates: Motion capture gives better data, but is labor-intensive and costly—plus, you can only record so many “templates” of movement.

Why Generative AI and Simulations Only Go So Far

Modern AI can learn to predict what comes next in video, but—

It still needs huge data sets, much larger than exist for grappling.
Simulators (physics engines) can’t capture every nuance of real life, making them brittle for coaching or fine-tuned feedback.

But there’s hope! Sports like golf and skiing have succeeded in modeling movement for analysis. Dr. Peter suggests judo will benefit as tech evolves, especially for position-based moves or solo analysis.

The Future: Can We Quantify “Kazushi?”

Imagine the tools:

You film your uchimata, and the tech spits out: “You need 2 more inches under the center of gravity.”
Olympian comparison: See, with lines and data, exactly why Ono’s osoto is special and how yours differs.

For coaching, commentary, and athlete development, these advances could eventually offer…

New ways to break down mistakes
Reinforcement for what feels right on the mat
Data-driven answers to decades-old judo questions (“How much kazushi is enough?”)

Takeaways from This Episode

The dream of a “Judo Bot” is inspiring but full of technical hurdles.
AI for grappling is real, but it’s tough—because movement is noisy, information is sparse, and human learning is still way ahead of digital models.
As tech improves, the judo world will benefit, but there’s no replacing feel, experience, or mat time—yet.

Final Thought:
Whether or not robots or AI ever master judo, the quest to understand our art more deeply—inch by inch—is here to stay. And that’s a win for coaches, athletes, and tech geeks alike.

Special Offer: FUJI GI’s 20% OFF

Upgrade your training with the same high-quality gis trusted by Shintaro Higashi! For a limited time, you can get 20% off all Fuji Sports judo gis when you shop through this exclusive link. Don’t miss your chance to save on top-tier gear—click the link and elevate your judo game today!

20% OFF click here