Skip to content

DrEureka

Automate sim to real robot design with LLMs

Visit official site

What DrEureka is

The method is described as an LLM driven approach that automates and speeds up sim to real design for robot skills.

What you can do with it

  • Generate reward functions for a target task
  • Produce parameters for domain randomisation
  • Test policies under varied simulated conditions

Method notes

  • Safety instructions shape the reward design
  • Physical priors are passed to the LLM
  • Trained policies are prepared for real deployment

Who it is for

  • Developers acquiring and deploying robot skills
  • Engineers automating robot task design and tests
  • Researchers in robotics and artificial intelligence

Pros & cons

✓ What we like

  • Safety instructions feed into reward design
  • Policies hold balance under terrain changes and disturbances

! What to watch out for

  • Task descriptions, safety instructions and environment source code are required
  • The workflow runs through several simulation stages

FAQ

What inputs are needed?

The steps name the task, safety instructions and environment source code.

Who is it described for?

The audience notes name robot skill developers, automation engineers and researchers.

Last reviewed: 2026-11-15

More AI coding tools tools

View all →

How we review