File size: 9,494 Bytes
825cff4 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 | # Overview
<p align="center">
<img width="24.0%" src="../images/task_lift.gif">
<img width="24.0%" src="../images/task_can.gif">
<img width="24.0%" src="../images/task_tool_hang.gif">
<img width="24.0%" src="../images/task_square.gif">
<img width="24.0%" src="../images/task_lift_real.gif">
<img width="24.0%" src="../images/task_can_real.gif">
<img width="24.0%" src="../images/task_tool_hang_real.gif">
<img width="24.0%" src="../images/task_transport.gif">
</p>
**robomimic** is a framework for robot learning from demonstration.
It offers a broad set of demonstration datasets collected on robot manipulation domains and offline learning algorithms to learn from these datasets.
**robomimic** aims to make robot learning broadly *accessible* and *reproducible*, allowing researchers and practitioners to benchmark tasks and algorithms fairly and to develop the next generation of robot learning algorithms.
## Core Features
<!-- <div style="box-sizing:border-box;" >
<section class="page-section" style="box-sizing:border-box;display:block;" >
<div class="container" style="box-sizing:border-box;width:100%;padding-right:0.75rem;padding-left:0.75rem;margin-right:auto;margin-left:auto;" >
<div class="row text-center" style="box-sizing:border-box;display:flex;flex-wrap:wrap;margin-right:-0.75rem;margin-left:-0.75rem;" >
<div class="col-lg-3 col-md-6" style="box-sizing:border-box;position:relative;width:100%;padding-right:0.75rem;padding-left:0.75rem;flex:0 0 50%;max-width:50%;" >
<div class="feature-box" style="box-sizing:border-box;background-attachment:scroll;padding-top:30px;padding-bottom:30px;padding-right:20px;padding-left:20px;margin-bottom:50px;text-align:center;border-width:1px;border-style:solid;border-color:#e6e6e6;height:400px;position:relative;background-color:#DEEEFB;background-image:none;background-repeat:repeat;background-position:top left;" >
<h3 style="box-sizing:border-box;margin-top:0;margin-bottom:0.5rem;" >Standardized Datasets</h3>
<p class="text-muted" style="box-sizing:border-box;margin-top:0;margin-bottom:1rem;" >Datasets collected from different sources (single proficient human, multiple humans, and machine-generated) across simulated and real-world tasks spanning multiple robots and environments</p>
</div>
</div>
<div class="col-lg-3 col-md-6" style="box-sizing:border-box;position:relative;width:100%;padding-right:0.75rem;padding-left:0.75rem;flex:0 0 50%;max-width:50%;" >
<div class="feature-box" style="box-sizing:border-box;background-attachment:scroll;padding-top:30px;padding-bottom:30px;padding-right:20px;padding-left:20px;margin-bottom:50px;text-align:center;border-width:1px;border-style:solid;border-color:#e6e6e6;height:400px;position:relative;background-color:#DEEEFB;background-image:none;background-repeat:repeat;background-position:top left;" >
<h3 style="box-sizing:border-box;margin-top:0;margin-bottom:0.5rem;" >Suite of Learning Algorithms</h3>
<p class="text-muted" style="box-sizing:border-box;margin-top:0;margin-bottom:1rem;" >High-quality implementations of offline learning algorithms, including BC, BC-RNN, HBC, IRIS, BCQ, CQL, and TD3-BC</p>
</div>
</div>
<div class="col-lg-3 col-md-6" style="box-sizing:border-box;position:relative;width:100%;padding-right:0.75rem;padding-left:0.75rem;flex:0 0 50%;max-width:50%;" >
<div class="feature-box" style="box-sizing:border-box;background-attachment:scroll;padding-top:30px;padding-bottom:30px;padding-right:20px;padding-left:20px;margin-bottom:50px;text-align:center;border-width:1px;border-style:solid;border-color:#e6e6e6;height:400px;position:relative;background-color:#DEEEFB;background-image:none;background-repeat:repeat;background-position:top left;" >
<h3 style="box-sizing:border-box;margin-top:0;margin-bottom:0.5rem;" >Modular Design</h3>
<p class="text-muted" style="box-sizing:border-box;margin-top:0;margin-bottom:1rem;" >Support for learning both low-dimensional and visuomotor policies, diverse network architectures, and interface to easily use external datasets</p>
</div>
</div>
<div class="col-lg-3 col-md-6" style="box-sizing:border-box;position:relative;width:100%;padding-right:0.75rem;padding-left:0.75rem;flex:0 0 50%;max-width:50%;" >
<div class="feature-box" style="box-sizing:border-box;background-attachment:scroll;padding-top:30px;padding-bottom:30px;padding-right:20px;padding-left:20px;margin-bottom:50px;text-align:center;border-width:1px;border-style:solid;border-color:#e6e6e6;height:400px;position:relative;background-color:#DEEEFB;background-image:none;background-repeat:repeat;background-position:top left;" >
<h3 style="box-sizing:border-box;margin-top:0;margin-bottom:0.5rem;" >Flexible Experiment Workflow</h3>
<p class="text-muted" style="box-sizing:border-box;margin-top:0;margin-bottom:1rem;" >Utilities for running hyperparameter sweeps, visualizing demonstration data and trained policies, and collecting new datasets using trained policies</p>
</div>
</div>
</div>
</div>
</section>
</div> -->
<p align="center">
<img style="width:100.0%;height:auto;" src="../images/core_features.png">
</p>
<!--
<style>
.column {
width: 45%;
float: left;
margin-right: 3%;
margin-bottom: 20px;
text-align: center;
padding: 20px;
height: 250px;
}
.column:last-child {
margin-right: 0;
}
.clear {
clear: both;
}
@media screen and (max-width : 1024px) {
.column {
width: 50%;
float: left;
margin-right: 0;
padding: 15px;
}
}
@media screen and (max-width : 767px) {
.column {
width: 100%;
float: none;
padding: 15px 0;
}
}
</style>
<div class="row">
<div class="column" style="background-color:#DEEEFB;">
<h4>Suite of Learning Algorithms</h4>
<p>High-quality implementations of offline learning algorithms, including BC, BC-RNN, HBC, IRIS, BCQ, CQL, and TD3-BC</p>
</div>
<div class="column" style="background-color:#DEEEFB;">
<h4>Standardized Datasets</h4>
<p>Datasets collected from different sources (single proficient human, multiple humans, and machine-generated) across simulated and real-world tasks spanning multiple robots and environments</p>
</div>
<div class="column" style="background-color:#DEEEFB;">
<h4>Modular Design</h4>
<p>Support for learning both low-dimensional and visuomotor policies, diverse network architectures, and interface to easily use external datasets</p>
</div>
<div class="column" style="background-color:#DEEEFB;">
<h4>Flexible Experiment Workflow</h4>
<p>Utilities for running hyperparameter sweeps, visualizing demonstration data and trained policies, and collecting new datasets using trained policies</p>
</div>
</div>
-->
<!-- 1. **Offline Learning Algorithms**
High-quality implementations of offline learning algorithms, including BC, BC-RNN, HBC, IRIS, BCQ, CQL, and TD3-BC
2. **Standardized Datasets**
Datasets collected from different sources (single proficient human, multiple humans, and machine-generated) across simulated and real-world tasks spanning multiple robots and environments
3. **Modular Design**
Support for learning both low-dimensional and visuomotor policies, diverse network architectures, and interface to easily use external datasets
4. **Flexible Experiment Workflow**
Utilities for running hyperparameter sweeps, visualizing demonstration data and trained policies, and collecting new datasets using trained policies -->
## Reproducing benchmarks
The robomimic framework also makes reproducing the results from different benchmarks and datasets easy. See the [datasets page](../datasets/overview.html) for more information on downloading datasets and reproducing experiments.
## Troubleshooting
Please see the [troubleshooting](../miscellaneous/troubleshooting.html) section for common fixes, or [submit an issue](https://github.com/ARISE-Initiative/robomimic/issues) on our github page.
## Contributing to robomimic
This project is part of the broader [Advancing Robot Intelligence through Simulated Environments (ARISE) Initiative](https://github.com/ARISE-Initiative), with the aim of lowering the barriers of entry for cutting-edge research at the intersection of AI and Robotics.
The project originally began development in late 2018 by researchers in the [Stanford Vision and Learning Lab](http://svl.stanford.edu/) (SVL).
Now it is actively maintained and used for robotics research projects across multiple labs.
We welcome community contributions to this project.
For details please check our [contributing guidelines](../miscellaneous/contributing.html).
## Citation
Please cite [this paper](https://arxiv.org/abs/2108.03298) if you use this framework in your work:
```bibtex
@inproceedings{robomimic2021,
title={What Matters in Learning from Offline Human Demonstrations for Robot Manipulation},
author={Ajay Mandlekar and Danfei Xu and Josiah Wong and Soroush Nasiriany and Chen Wang and Rohun Kulkarni and Li Fei-Fei and Silvio Savarese and Yuke Zhu and Roberto Mart\'{i}n-Mart\'{i}n},
booktitle={Conference on Robot Learning (CoRL)},
year={2021}
}
```
|