# mint-sjtu/evo-rl architecture

> Architecture diagram of the GitHub repository mint-sjtu/evo-rl, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2026-10-04). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/mint-sjtu/evo-rl
- Repository: https://github.com/mint-sjtu/evo-rl

## Overview

Evo-RL is a LeRobot-based toolkit for real-world robot learning. Its documented workflow runs from hardware setup and demonstration collection through value-function training and inference, policy training, and closed-loop rollout for another data-collection round. The graph includes that workflow, shared policy/processor infrastructure, and notable policy and robot-runtime capabilities. Workflow links follow the README; source-index references support only the sampled internal links, so unsampled implementation details are not inferred.

Main entry points: [`src/lerobot/scripts/lerobot_train.py`](https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/scripts/lerobot_train.py) (Policy Training), [`src/lerobot/datasets/lerobot_dataset.py`](https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/datasets/lerobot_dataset.py) (Robot Dataset), [`src/lerobot/robots/robot.py`](https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/robots/robot.py) (Robot Hardware), [`src/lerobot/policies/factory.py`](https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/policies/factory.py) (Policy Factory)

## Diagram (Mermaid)

```mermaid
flowchart TD

subgraph group_collection["Robot Data Collection"]
  node_robot["Robot Hardware<br/>[robot.py]"]
  node_teleop["Teleoperation<br/>[teleoperator.py]"]
  node_camera["Camera Capture<br/>[camera.py]"]
  node_record["Data Recording<br/>[lerobot_record.py]"]
  node_dataset[("Robot Dataset<br/>[lerobot_dataset.py]")]
  node_subtask["Subtask Annotation"]
end

subgraph group_learning["Value And Policy Learning"]
  node_value_train["Value Training"]
  node_value_model["Value Model"]
  node_value_infer["Value Inference"]
  node_policy_train["Policy Training<br/>[lerobot_train.py]"]
end

subgraph group_runtime["Policy Runtime"]
  node_policy_factory["Policy Factory<br/>[factory.py]"]
  node_policy_models["Policy Models<br/>[pretrained.py]"]
  node_wallx["Wall-X Policy<br/>[modeling_wall_x.py]"]
  node_qwen["Qwen Vision Model<br/>[qwen2_5_vl_moe.py]"]
  node_rtc["Realtime Control<br/>[modeling_rtc.py]"]
  node_eval["Policy Evaluation<br/>[lerobot_eval.py]"]
  node_teleoperate["Robot Teleoperation"]
  node_async_client["Async Robot Client<br/>[robot_client.py]"]
end

subgraph group_foundation["Shared LeRobot Foundation"]
  node_processors["Data Processors<br/>[pipeline.py]"]
  node_converters["Data Converters<br/>[converters.py]"]
  node_policy_config["Policy Configuration<br/>[policies.py]"]
  node_optimizer["Optimizers<br/>[optimizers.py]"]
  node_scheduler["Schedulers<br/>[schedulers.py]"]
  node_hub["Hub Utilities<br/>[hub.py]"]
end

node_operator(("Robot Operator"))

node_operator -->|"operates"| node_teleop
node_teleop -->|"controls"| node_robot
node_camera -->|"captures frames"| node_record
node_robot -->|"provides observations"| node_record
node_record -->|"records episodes"| node_dataset
node_dataset -->|"trains from"| node_value_train
node_dataset -->|"trains from"| node_policy_train
node_dataset -->|"annotates"| node_subtask
node_value_train -->|"trains"| node_value_model
node_value_model -->|"scores episodes"| node_value_infer
node_value_infer -->|"informs"| node_policy_train
node_policy_train -->|"configures policy"| node_policy_factory
node_policy_factory -->|"loads configuration"| node_policy_config
node_policy_factory -->|"creates model"| node_policy_models
node_policy_train -.->|"uses"| node_optimizer
node_policy_train -.->|"uses"| node_scheduler
node_wallx -->|"uses vision model"| node_qwen
node_wallx -.->|"supports realtime control"| node_rtc
node_policy_models -->|"publishes models"| node_hub
node_policy_train -.->|"processes policy data"| node_processors
node_processors -->|"uses converters"| node_converters
node_eval -.->|"loads policy"| node_policy_factory
node_eval -->|"runs rollout"| node_robot
node_eval -->|"records next round"| node_dataset
node_teleoperate -->|"drives hardware"| node_robot
node_async_client -->|"sends actions"| node_robot

click node_robot "https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/robots/robot.py"
click node_teleop "https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/teleoperators/teleoperator.py"
click node_camera "https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/cameras/camera.py"
click node_record "https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/scripts/lerobot_record.py"
click node_dataset "https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/datasets/lerobot_dataset.py"
click node_subtask "https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/data_processing/sarm_annotations/subtask_annotation.py"
click node_value_train "https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/scripts/lerobot_value_train.py"
click node_value_model "https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/values/pistar06/modeling_pistar06.py"
click node_value_infer "https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/scripts/lerobot_value_infer.py"
click node_policy_train "https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/scripts/lerobot_train.py"
click node_policy_factory "https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/policies/factory.py"
click node_policy_models "https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/policies/pretrained.py"
click node_wallx "https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/policies/wall_x/modeling_wall_x.py"
click node_qwen "https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/policies/wall_x/qwen_model/qwen2_5_vl_moe.py"
click node_rtc "https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/policies/rtc/modeling_rtc.py"
click node_eval "https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/scripts/lerobot_eval.py"
click node_teleoperate "https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/scripts/lerobot_teleoperate.py"
click node_async_client "https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/async_inference/robot_client.py"
click node_processors "https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/processor/pipeline.py"
click node_converters "https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/processor/converters.py"
click node_policy_config "https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/configs/policies.py"
click node_optimizer "https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/optim/optimizers.py"
click node_scheduler "https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/optim/schedulers.py"
click node_hub "https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/utils/hub.py"

classDef toneNeutral fill:#f8fafc,stroke:#334155,stroke-width:1.5px,color:#0f172a
classDef toneBlue fill:#dbeafe,stroke:#2563eb,stroke-width:1.5px,color:#172554
classDef toneAmber fill:#fef3c7,stroke:#d97706,stroke-width:1.5px,color:#78350f
classDef toneMint fill:#dcfce7,stroke:#16a34a,stroke-width:1.5px,color:#14532d
classDef toneRose fill:#ffe4e6,stroke:#e11d48,stroke-width:1.5px,color:#881337
classDef toneIndigo fill:#e0e7ff,stroke:#4f46e5,stroke-width:1.5px,color:#312e81
classDef toneTeal fill:#ccfbf1,stroke:#0f766e,stroke-width:1.5px,color:#134e4a
class node_robot,node_teleop,node_camera,node_record,node_dataset,node_subtask toneBlue
class node_value_train,node_value_model,node_value_infer,node_policy_train toneAmber
class node_policy_factory,node_policy_models,node_wallx,node_qwen,node_rtc,node_eval,node_teleoperate,node_async_client toneMint
class node_processors,node_converters,node_policy_config,node_optimizer,node_scheduler,node_hub toneRose
class node_operator toneIndigo
```

## Components

### Robot Data Collection

- [Robot Hardware](https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/robots/robot.py): `src/lerobot/robots/robot.py`
- [Teleoperation](https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/teleoperators/teleoperator.py): `src/lerobot/teleoperators/teleoperator.py`
- [Camera Capture](https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/cameras/camera.py): `src/lerobot/cameras/camera.py`
- [Data Recording](https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/scripts/lerobot_record.py): `src/lerobot/scripts/lerobot_record.py`
- [Robot Dataset](https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/datasets/lerobot_dataset.py): `src/lerobot/datasets/lerobot_dataset.py`
- [Subtask Annotation](https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/data_processing/sarm_annotations/subtask_annotation.py): `src/lerobot/data_processing/sarm_annotations/subtask_annotation.py`

### Value And Policy Learning

- [Value Training](https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/scripts/lerobot_value_train.py): `src/lerobot/scripts/lerobot_value_train.py`
- [Value Model](https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/values/pistar06/modeling_pistar06.py): `src/lerobot/values/pistar06/modeling_pistar06.py`
- [Value Inference](https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/scripts/lerobot_value_infer.py): `src/lerobot/scripts/lerobot_value_infer.py`
- [Policy Training](https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/scripts/lerobot_train.py): `src/lerobot/scripts/lerobot_train.py`

### Policy Runtime

- [Policy Factory](https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/policies/factory.py): `src/lerobot/policies/factory.py`
- [Policy Models](https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/policies/pretrained.py): `src/lerobot/policies/pretrained.py`
- [Wall-X Policy](https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/policies/wall_x/modeling_wall_x.py): `src/lerobot/policies/wall_x/modeling_wall_x.py`
- [Qwen Vision Model](https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/policies/wall_x/qwen_model/qwen2_5_vl_moe.py): `src/lerobot/policies/wall_x/qwen_model/qwen2_5_vl_moe.py`
- [Realtime Control](https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/policies/rtc/modeling_rtc.py): `src/lerobot/policies/rtc/modeling_rtc.py`
- [Policy Evaluation](https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/scripts/lerobot_eval.py): `src/lerobot/scripts/lerobot_eval.py`
- [Robot Teleoperation](https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/scripts/lerobot_teleoperate.py): `src/lerobot/scripts/lerobot_teleoperate.py`
- [Async Robot Client](https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/async_inference/robot_client.py): `src/lerobot/async_inference/robot_client.py`

### Shared LeRobot Foundation

- [Data Processors](https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/processor/pipeline.py): `src/lerobot/processor/pipeline.py`
- [Data Converters](https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/processor/converters.py): `src/lerobot/processor/converters.py`
- [Policy Configuration](https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/configs/policies.py): `src/lerobot/configs/policies.py`
- [Optimizers](https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/optim/optimizers.py): `src/lerobot/optim/optimizers.py`
- [Schedulers](https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/optim/schedulers.py): `src/lerobot/optim/schedulers.py`
- [Hub Utilities](https://github.com/mint-sjtu/evo-rl/blob/main/src/lerobot/utils/hub.py): `src/lerobot/utils/hub.py`

### Other components

- **Robot Operator**

## Connections

- Robot Operator → Teleoperation: operates (evidence: `README.md`)
- Teleoperation → Robot Hardware: controls (evidence: `README.md`)
- Camera Capture → Data Recording: captures frames (evidence: `README.md`)
- Robot Hardware → Data Recording: provides observations (evidence: `README.md`)
- Data Recording → Robot Dataset: records episodes (evidence: `README.md`)
- Robot Dataset → Value Training: trains from (evidence: `README.md`)
- Robot Dataset → Policy Training: trains from (evidence: `README.md`)
- Robot Dataset → Subtask Annotation: annotates (evidence: `src/lerobot/data_processing/sarm_annotations/subtask_annotation.py`)
- Value Training → Value Model: trains (evidence: `README.md`)
- Value Model → Value Inference: scores episodes (evidence: `README.md`)
- Value Inference → Policy Training: informs (evidence: `README.md`)
- Policy Training → Policy Factory: configures policy (evidence: `src/lerobot/policies/factory.py`)
- Policy Factory → Policy Configuration: loads configuration (evidence: `src/lerobot/policies/factory.py`)
- Policy Factory → Policy Models: creates model (evidence: `src/lerobot/policies/factory.py`)
- Policy Training → Optimizers: uses
- Policy Training → Schedulers: uses
- Wall-X Policy → Qwen Vision Model: uses vision model (evidence: `src/lerobot/policies/wall_x/modeling_wall_x.py`)
- Wall-X Policy → Realtime Control: supports realtime control (evidence: `src/lerobot/policies/rtc/modeling_rtc.py`)
- Policy Models → Hub Utilities: publishes models (evidence: `src/lerobot/policies/pretrained.py`)
- Policy Training → Data Processors: processes policy data
- Data Processors → Data Converters: uses converters (evidence: `src/lerobot/processor/pipeline.py`)
- Policy Evaluation → Policy Factory: loads policy
- Policy Evaluation → Robot Hardware: runs rollout (evidence: `README.md`)
- Policy Evaluation → Robot Dataset: records next round (evidence: `README.md`)
- Robot Teleoperation → Robot Hardware: drives hardware (evidence: `README.md`)
- Async Robot Client → Robot Hardware: sends actions (evidence: `src/lerobot/async_inference/robot_client.py`)
