Key Findings
NTT has developed a physical AI technology that allows robots to simulate multiple future outcomes before executing an action, selecting the path with the lowest probability of failure. In a practical application involving the stacking of items by a small robot, the success rate improved from 73.1% to 92.3% by evaluating five different action candidates and avoiding those leading to object drops.
Technical Details
The system utilizes a world model that combines a video generation model with a proprietary “physical control adapter.” While standard video generation models can produce “plausible” visuals that may violate physics (such as objects disappearing), the physical control adapter ensures accuracy by:
- Estimating physical quantities including position, posture, velocity, angular velocity, mass, and moment of inertia.
- Calculating future states based on laws of physics and equations of motion.
- Using these calculated states as conditions for the video generation model.
In the “OBJ3D” benchmark for physical prediction accuracy, NTT’s model achieved an LPIPS×100 score of 7.81, outperforming NVIDIA Cosmos Predict 2.5 (21.90) and the existing “OCK” method (7.92).
Background and Context
Traditional industrial robots excel in controlled environments with fixed programs but struggle with variable objects. Recent Vision-Language-Action (VLA) models improve versatility, but success rates remain low for tasks involving diverse shapes, weights, and friction—such as airport luggage handling—where one movement can cause a chain reaction of collapses.
Significance and Outlook
NTT intends to implement this technology first in airport baggage loading and unloading, collaborating with Mujin to integrate the world model with Mujin’s robot control technology. A primary remaining challenge is processing speed; currently, predicting a 5-second future takes 2 to 3 seconds. NTT plans to utilize model distillation to reduce latency and ensure the robot can operate without significant pauses.
Source: https://news.mynavi.jp/techplus/article/20261111-5296066/
Get our weekly technology intelligence — free
Receive an infographic that lets you judge at a glance whether each field’s analysis report is worth reading.
Subscribe Free — Weekly Tech Intelligence
By subscribing, you’ll receive Troy-Technical’s weekly technology intelligence newsletter.
- Your email and selected fields are used only to deliver the newsletter.
- We never share your information with third parties.
- You can unsubscribe anytime via the link in each email.
See our Privacy Policy for details.
Takes about a minute · Unsubscribe anytime



