# New data‑collection method boosts robot world‑model control

> Action Influence Maximization reduces training data needed for controllable video models by 85×, researchers say.

Oossa · 2026-10-21 · https://oossa.com/en/new-data-collection-method-boosts-robot-world-model-control

Researchers led by Samuel Bateman introduced Action Influence Maximization (AIM), an algorithm that actively gathers training episodes to make video‑based world models pay attention to commanded actions. In tests on simulated navigation and pixel‑based manipulation, AIM added just 500 new episodes to an existing set of 3,000 expert demos and matched the controllability of a model trained on 300,000 expert episodes. The method outperformed standard exploration and intrinsic‑motivation techniques in data efficiency.

## The facts

- AIM achieved an 85× reduction in data, needing 500 episodes versus 300k expert demos.
- Paper posted on arXiv on 2026‑10‑21.

## Why it matters

For robot developers, AIM means they can train more responsive control models with far fewer costly demonstration recordings.

## Sources & references

1. [Training Controllable World Models via Action Influence Maximization](https://arxiv.org/abs/2610.27387) – arXiv cs.RO, 2026-10-21

Last updated: 2026-10-21