← Back to the feed
Robotics & manufacturingResearch findings

CoDA · Co-evolution Framework for Data Engines and Action Policies

Original published 2026.10.16 02:59 KSTAdded to Blackdeck 2026.10.18 13:44 KST

What has changed?

A simulation data engine and action policy co-evolve using shared memory of failures and solutions.

Key points

  1. Simulation environments adapt based on policy failure diagnoses.
  2. A shared memory enables mutual improvement between data engine and policy.
  3. Out-of-distribution success improves from 5.3% to 27.6% across four manipulation tasks.

Connect it to your work

Failure cases are diagnosed and used to generate environments and training programs that improve policy performance.

What remains unproven

The abstract reports simulation results; real-world performance is not established by this input.

Originals used for this note

Links open the original publisher in a new tab.

  1. arxivCoDA: Co-Evolution of Data Engines and Action Policies