sim2real-transfer

Use when a policy or controller trained/tuned in simulation must run on the real robot — close the gap with domain randomization for what you can't measure and system identification for what you can.

Files1

View on GitHub

SKILL.md

GitHub

When to use. Before any sim-trained behavior touches hardware. Why: the sim-to-real gap fails policies silently; randomize what you can't identify, identify what you can.

Procedure.

  1. System-ID measurable params (joint friction, rotor inertia, actuator delay) against real logs so the nominal sim is correct.
  2. Domain-randomize the rest (mass, friction, latency, lighting, textures) — tune ranges (too wide is unlearnable, too narrow overfits).
  3. Validate across engines (dual-sim) before hardware.
  4. Done when: the policy holds up in a held-out randomized sim and a supervised real-robot trial, with the gap quantified.

Knowledge: KNOWLEDGE-LIBRARY.md §B3.

Related in other

limits-colimits

Context management for Claude Code. Hooks maintain state via ledgers and handoffs. MCP execution without context polluti...

by parcadei
3,392

ddttom-allaboutv2-claude-skills-block-collection-and-party

The Block Collection and Block Party are repositories for existing AEM blocks, build tools, code snippets, integration p...

by mattnigh
20

nodnarbnitram-claude-code-extensions-claude-skills-esphome-confi

This skill should be used when the user asks to "create an esphome config", "set up an esp32 device", "configure esphome...

by mattnigh
20