Robot Platforms
0
Continuous Open-Source Real-World RL
A LeRobot-aligned, full-pipeline real-world offline RL workflow for real robots: collect rollout data, train value functions and policies offline, infer advantage tags, and iterate deployment in a closed loop.
pipeline.sh
lerobot-human-inloop-record
--dataset.repo_id=<DATASET>
lerobot-value-train
--value.type=pistar06
lerobot-value-infer
--acp.positive_ratio=0.3
lerobot-train
--acp.enable=true
Robot Platforms
0
Open Task Datasets
0
Open Models
0
Release Year
0
Project Focus
Runnable workflows on SO101 and AgileX PiPER covering collection, offline value/policy training, and iterative deployment.
Codebase, model layout, dataset structure, and command-line templates are released to reduce setup friction and improve reproducibility.
Reproduce existing real-world RL methods, extend value/policy variants, and continuously publish experiments and benchmarks for collective progress.
Method
System preparation to ensure robot, sensors, and control stack are ready.
Interactive data collection with human guidance and episode-level outcomes.
Value modeling to estimate trajectory quality from accumulated data.
Signal annotation to convert value outputs into usable training supervision.
Policy update and deployment to run the improved policy in real tasks.
Data refresh and iteration to continue improving robustness and success rate.
Results
Illustrative trend view for project webpage. Replace with your run logs for paper-ready plots.
Current default ACP ratio in README pipeline: --acp.positive_ratio=0.3.
Qualitative Demos
VALUE VISUAL
Frame-level value visualization on real robot rollouts.
POLICY ROLLOUT
Final policy behavior clips in target task execution.
HUMAN-IN-THE-LOOP
Human guidance segments and autonomous recovery outcomes.
Architecture
Evo-RL reuses LeRobot's practical data and inference stack while adding value-guided filtering signals (`value`, `advantage`, `acp_indicator`) to improve iterative policy training.
The same design supports single-GPU and accelerate-based multi-GPU workflows for both value and policy stages.
Reproducibility
lerobot-human-inloop-record \
--robot.type=<ROBOT_TYPE> \
--teleop.type=<TELEOP_TYPE> \
--dataset.repo_id=<HF_USER_OR_ORG/DATASET> \
--dataset.single_task="<TASK_TEXT>" \
--dataset.num_episodes=<NUM_EPISODES> \
--display_data=true
lerobot-value-train \
--dataset.repo_id=<HF_USER_OR_ORG/DATASET> \
--value.type=<VALUE_TYPE> \
--batch_size=<BATCH_SIZE> \
--output_dir=outputs/value_train/<RUN_NAME> \
--job_name=<RUN_NAME>
lerobot-train \
--dataset.repo_id=<HF_USER_OR_ORG/DATASET> \
--policy.type=<POLICY_TYPE> \
--steps=<TRAIN_STEPS> \
--acp.enable=true \
--acp.indicator_dropout_prob=0.3 \
--output_dir=outputs/train/<RUN_NAME>
Model & Community
Scan to join EvoMind's WeChat community and follow project updates.
Contact business@evomind-tech.com
Citation
@misc{evorl2026,
title = {Evo-RL: Continuous Open-Source Real-World RL on SO101 and Beyond},
author = {Evo-RL Contributors},
year = {2026},
howpublished = {\url{https://github.com/MINT-SJTU/Evo-RL}}
}