robotwin-instruction-to-env

To generate a RoboTwin environment from a task description, provide a task name and brief to normalize scene requirements, write conservative setup code, and render a deterministic first frame. The process reuses nearby env patterns and local pose priors to ensure valid default orientations.

First-frame rendering for robot simulation tasks validates generated environment code by producing a deterministic initial scene image. It uses the repository's render path and local pose manuals to verify layout relations, camera constraints, and safe defaults before downstream testing.

Repo discovery for robot simulation environments inspects the local repository structure to validate compatibility and reuse nearby env patterns. It loads local asset pose priors to ground default orientations, ensuring the generated envs/<task>.py file follows existing conventions.

Yes, you can use local asset pose priors to set default orientations in RoboTwin. The skill loads these local pose manuals to ground object placements instead of guessing, ensuring stable initialization for downstream testing or demonstrations.

The limitation of instruction-driven scene creation for RoboTwin is its strict dependency on a specific local repo layout. Without the required envs/, task_config/, references/ directories and a render path, the skill cannot validate generated frames or guarantee safe defaults.