# RoboCasa (/docs/evaluation/benchmark-hub/robocasa)



## About [#about]

RoboCasa external benchmark task manifest for long-horizon robot manipulation policy rollouts.

## Metrics [#metrics]

RoboCasa external benchmark task manifest for long-horizon robot manipulation policy rollouts.

### Task YAML [#task-yaml]

`worldfoundry/data/benchmarks/tasks/external/robocasa.yaml`

### Primary [#primary]

* `success_rate`

| Metric            | Focus                                                                                        |
| ----------------- | -------------------------------------------------------------------------------------------- |
| `success_rate`    | Share of episodes or tasks that complete successfully.                                       |
| `task_success`    | Share of episodes or tasks that complete successfully.                                       |
| `episode_success` | Share of episodes or tasks that complete successfully.                                       |
| `reward`          | Cumulative reward on RoboCasa kitchen manipulation tasks in the official domestic simulator. |

## Environment setup [#environment-setup]

### Runtime environment [#runtime-environment]

* **Conda env:** `robocasa-headless-venv`.
* **Separate setup:** defaults to the unified WorldFoundry env (`needs_new_env: false`); a benchmark-only conda env is usually not required.
* RoboCasa/robosuite import 已在 headless venv 可见；完整官方状态仍需要 assets、task suite 和 policy rollout。

### Evaluation data [#evaluation-data]

* RoboCasa uses simulator assets and generated policy rollouts rather than a single public HF benchmark table.
* Candidate model outputs: set `WORLDFOUNDRY_ROBOCASA_RESULTS_PATH` to the generated-video or rollout artifact root.
* `${WORLDFOUNDRY_CACHE_DIR}/repos/robocasa--robocasa`

### Checkpoints & assets [#checkpoints--assets]

* This benchmark is primarily rollout- or official-result-driven; see catalog `requires` and the task YAML for policy/simulator checkpoints.

### Key environment variables [#key-environment-variables]

* `WORLDFOUNDRY_ROBOCASA_ROOT`
* `WORLDFOUNDRY_ROBOCASA_RESULTS_PATH`

### Verify setup [#verify-setup]

```bash
PYTHONPATH=${WORLDFOUNDRY_CACHE_DIR}/repos/robocasa--robocasa:src ${WORLDFOUNDRY_CONDA_ENVS_ROOT}/worldfoundry-robocasa-headless-venv/bin/python -m worldfoundry.evaluation.tasks.execution.orchestration.runtime_preflight --profile robocasa --manifest worldfoundry/data/benchmarks/runtime_profiles/official/robocasa.yaml --output-dir tmp/worldfoundry_preflight/robocasa --json
```

## Run Evaluation [#run-evaluation]

After preparing the assets and candidate artifacts listed on this page, use the public run entry points below.

### Set Candidate Artifacts [#set-candidate-artifacts]

```bash
export WORLDFOUNDRY_GENERATED_ARTIFACT_DIR=/path/to/generated/artifacts
```

### Import Official Results [#import-official-results]

```bash
worldfoundry-eval zoo benchmark-run \
  --benchmark-id robocasa \
  --mode normalizer \
  --official-results-path /path/to/official/results-or-report \
  --generated-artifact-dir "${WORLDFOUNDRY_GENERATED_ARTIFACT_DIR}" \
  --output-dir tmp/robocasa/normalizer \
  --json
```

### Score Generated Artifacts [#score-generated-artifacts]

```bash
worldfoundry-eval zoo benchmark-run \
  --benchmark-id robocasa \
  --mode official-run \
  --generated-artifact-dir "${WORLDFOUNDRY_GENERATED_ARTIFACT_DIR}" \
  --output-dir tmp/robocasa/official-run \
  --json
```

### Direct In-Tree Runner [#direct-in-tree-runner]

```bash
worldfoundry-eval zoo benchmark-run --benchmark-id robocasa --mode official-validation --official-results-path <official_results.json> --output-dir <out> --json
```

## Requirements [#requirements]

### Inputs [#inputs]

* Official RoboCasa repository checkout
* RoboCasa simulator assets
* Long-horizon policy checkpoint
* Official task and scene split declaration

### Environment [#environment]

* `WORLDFOUNDRY_ROBOCASA_RESULTS_PATH`

### Outputs [#outputs]

* `scorecard.json`
* `raw_results.jsonl`

[← Benchmark Hub](/docs/evaluation/benchmark-hub)