Repository navigation
Expand file tree
/
Copy pathprojects.html
More file actions
154 lines (140 loc) · 9.98 KB
/
Copy pathprojects.html
File metadata and controls
154 lines (140 loc) · 9.98 KB
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
154
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.1//EN"
"http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en">
<head>
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<meta http-equiv="Content-Type" content="text/html;charset=utf-8" />
<link rel="icon" type="image/svg+xml" href="./images/SJTU-logo.svg">
<link rel="stylesheet" href="style.css" type="text/css" />
<title>Research & Projects | SAIL Lab @ SJTU</title>
<!-- MathJax -->
<script src='https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.5/latest.js?config=TeX-MML-AM_CHTML' async>
</script>
<script type="text/x-mathjax-config">
MathJax.Hub.Config({
TeX: { equationNumbers: { autoNumber: "AMS" } }
});
</script>
<!-- End MathJax -->
</head>
<body>
<header class="site-header">
<div class="brand-bar">
<div class="brand-inner">
<a class="brand-seal" href="https://en.sjtu.edu.cn/" target="_blank"><img src="./images/SJTU-logo-white.svg" alt="Shanghai Jiao Tong University" /></a>
<div class="brand-title">
<a href="index.html">SAIL Lab</a>
<span class="brand-sub">Safe AI and Robot Learning Lab · Shanghai Jiao Tong University</span>
</div>
<div class="brand-links">
<a href="https://github.com/SAIL-Research-Lab" target="_blank" title="SAIL Lab GitHub"><svg viewBox="0 0 16 16" aria-hidden="true"><path fill-rule="evenodd" d="M8 0C3.58 0 0 3.58 0 8c0 3.54 2.29 6.53 5.47 7.59.4.07.55-.17.55-.38 0-.19-.01-.82-.01-1.49-2.01.37-2.53-.49-2.69-.94-.09-.23-.48-.94-.82-1.13-.28-.15-.68-.52-.01-.53.63-.01 1.08.58 1.23.82.72 1.21 1.87.87 2.33.66.07-.52.28-.87.51-1.07-1.78-.2-3.64-.89-3.64-3.95 0-.87.31-1.59.82-2.15-.08-.2-.36-1.02.08-2.12 0 0 .67-.21 2.2.82.64-.18 1.32-.27 2-.27.68 0 1.36.09 2 .27 1.53-1.04 2.2-.82 2.2-.82.44 1.1.16 1.92.08 2.12.51.56.82 1.27.82 2.15 0 3.07-1.87 3.75-3.65 3.95.29.25.54.73.54 1.48 0 1.07-.01 1.93-.01 2.2 0 .21.15.46.55.38A8.01 8.01 0 0 0 16 8c0-4.42-3.58-8-8-8z"/></svg></a>
<a href="https://www.youtube.com/@ai-agent-research" target="_blank" title="SAIL Lab YouTube"><svg viewBox="0 0 24 24" aria-hidden="true"><path d="M23.5 6.19a3.02 3.02 0 0 0-2.12-2.14C19.5 3.55 12 3.55 12 3.55s-7.5 0-9.38.5A3.02 3.02 0 0 0 .5 6.19C0 8.07 0 12 0 12s0 3.93.5 5.81a3.02 3.02 0 0 0 2.12 2.14c1.88.5 9.38.5 9.38.5s7.5 0 9.38-.5a3.02 3.02 0 0 0 2.12-2.14C24 15.93 24 12 24 12s0-3.93-.5-5.81zM9.55 15.57V8.43L15.82 12l-6.27 3.57z"/></svg></a>
<a href="mailto:shangding.gu@sjtu.edu.cn">shangding.gu@sjtu.edu.cn</a>
</div>
<button id="nav-toggle" aria-label="Menu" onclick="document.body.classList.toggle('nav-open')">☰</button>
</div>
</div>
<nav class="site-nav">
<div class="nav-inner">
<ul class="nav-list">
<li class="nav-item"><a href="index.html">Home</a></li>
<li class="nav-item"><a href="group.html">People</a></li>
<li class="nav-item"><a href="publication.html">Publications</a></li>
<li class="nav-item current"><a href="projects.html">Research & Projects</a></li>
<li class="nav-item"><a href="softwares.html">Products & Software</a></li>
<li class="nav-item"><a href="tutorial_talk.html">Tutorials & Talks</a></li>
<li class="nav-item"><a href="academic_service.html">Academic Service</a></li>
<li class="nav-item"><a href="resources.html">Resources</a></li>
<li class="nav-item"><a href="join.html">Join Us</a></li>
</ul>
</div>
</nav>
</header>
<div class="page">
<h1>Research & Projects</h1>
<div class="muted" style="margin-top:-6px;">Selected research topics and systems developed in the lab (safe RL, robot learning, LLM agents, and beyond).</div>
<h2>Research</h2>
<p style="font-size: 17px;">
Concretely, the lab builds <b>algorithms, theory, benchmarks, and open-source systems</b> for safe decision-making under
uncertainty — spanning safe reinforcement learning, robot learning, and LLM-based agents, and evaluates them in robotics,
autonomous driving, smart grid, and semiconductor-manufacturing scenarios.
</p>
<p>
His work has been featured in leading publications, including top-tier journals and conferences such as
Science Advances, the Journal of Artificial Intelligence, IEEE Transactions on Pattern Analysis and Machine Intelligence, NeurIPS, and other prestigious venues.
</p>
<p>
<span class="chip">Safe Reinforcement Learning</span>
<span class="chip">Robust Reinforcement Learning</span>
<span class="chip">Robot Learning</span>
<span class="chip">Large Language Models & Agents</span>
<span class="chip">Multi-Agent Reinforcement Learning</span>
<span class="chip">AI Safety & Trustworthy AI</span>
<span class="chip">Semiconductor Manufacturing</span>
</p>
<p class="muted">More on our <a href="publication.html">publications</a> and <a href="softwares.html">open-source softwares</a>.</p>
<h2>Projects</h2>
<h2>2026</h2>
<div class="proj">
<img src="./images/projects/cheetahclaws_web_demo.gif" alt="CheetahClaws demo" />
<div>
<h3>CheetahClaws: An Agent Harness Infrastructure for Long-Horizon, Multi-Model, and Tool-Using AI Systems</h3>
<p>CheetahClaws is a fast and easy-to-use agent harness infrastructure designed for long-horizon, multi-model, and tool-using AI systems. It provides a flexible and extensible framework supporting frontier and local models, tool use, memory, multi-agent orchestration, and autonomous task execution for building and studying advanced AI agents.</p>
<p class="links"><strong>[</strong><a href="https://arxiv.org/pdf/2605.26112" target="_blank">Paper</a><strong>], [</strong><a href="https://github.com/SafeRL-Lab/cheetahclaws" target="_blank">Code</a><strong>]</strong></p>
</div>
</div>
<h2>2025</h2>
<div class="proj">
<img src="./images/projects/2024-gif-edit-overview.gif" alt="Robust Gymnasium overview" />
<div>
<h3>Robust Gymnasium: A Unified Modular Benchmark for Robust Reinforcement Learning</h3>
<p>This benchmark aims to advance robust reinforcement learning for real-world applications and domain adaptation. The benchmark provides a comprehensive set of tasks that cover various robustness requirements in the face of uncertainty on state, action, reward, and environmental dynamics, and spans diverse applications including control, robot manipulation, dexterous hands, and so on.</p>
<p class="links"><strong>[</strong><a href="https://github.com/SafeRL-Lab/Robust-Gymnasium/blob/main/docs/_static/paper/Robust_RL_benchmark-git.pdf" target="_blank">Paper</a><strong>], [</strong><a href="https://robust-rl.com/" target="_blank">Code</a><strong>]</strong></p>
</div>
</div>
<h2>2024</h2>
<div class="proj">
<img src="./images/projects/2024-teams-rl-overview-framework.png" alt="TeaMs-RL framework" />
<div>
<h3>TeaMs-RL: Teaching LLMs to Teach Themselves Better Instructions via Reinforcement Learning</h3>
<p>This work presents TeaMs-RL, a method that leverages reinforcement learning to directly generate instruction datasets for fine-tuning LLMs. By minimizing reliance on human feedback and external queries, TeaMs-RL enhances data quality, privacy, and model capabilities, making it a valuable approach for foundation-model post-training.</p>
<p class="links"><strong>[</strong><a href="https://arxiv.org/abs/2403.08694" target="_blank">Paper</a><strong>], [</strong><a href="https://github.com/SafeRL-Lab/TeaMs-RL" target="_blank">Code</a><strong>]</strong></p>
</div>
</div>
<h2>2024</h2>
<div class="proj">
<img src="./images/projects/2024-safe-marl-roundabout.gif" alt="Safe MARL roundabout scenario" />
<div>
<h3>Safe Multi-Agent Reinforcement Learning with Bilevel Optimization in Autonomous Driving</h3>
<p>This study introduces a safe MARL method based on a Stackelberg model with bi-level optimization, featuring two algorithms, CSQ and CS-MADDPG, for autonomous driving applications. Experimental results demonstrate its superior safety and performance compared with strong MARL baselines in challenging driving scenarios.</p>
<p class="links"><strong>[</strong><a href="https://arxiv.org/abs/2405.18209" target="_blank">Paper</a><strong>], [</strong><a href="https://github.com/SafeRL-Lab/Safe-MARL-in-Autonomous-Driving" target="_blank">Code</a><strong>]</strong></p>
</div>
</div>
<h2>2023</h2>
<div class="proj">
<img src="./images/projects/2023-safe-multi-robot-01.gif" alt="Safe multi-robot control" />
<div>
<h3>Safe Multi-Agent Reinforcement Learning for Multi-Robot Control</h3>
<p>We investigate safe MARL for multi-robot control on cooperative tasks, in which each individual robot has to not only meet its own safety constraints while maximizing its reward, but also consider those of others to guarantee safe team behaviors.</p>
<p class="links"><strong>[</strong><a href="https://sites.google.com/view/aij-safe-marl/" target="_blank">Paper</a><strong>], [</strong><a href="https://sites.google.com/view/aij-safe-marl/" target="_blank">Code</a><strong>]</strong></p>
</div>
</div>
<p class="muted">More topics we work on — including safe RL for semiconductor manufacturing and smart-grid operation, robust RL, offline RL, and agentic web research — are summarized on the <a href="index.html">homepage</a> and in our <a href="publication.html">publications</a>.</p>
</div>
<footer class="site-footer">
<div class="footer-inner">
© 2026 SAIL Lab (Safe AI and Robot Learning Lab) — School of Computer Science, Shanghai Jiao Tong University<br/>
Principal Investigator: Prof. Shangding Gu (顾尚定) ·
<a href="mailto:shangding.gu@sjtu.edu.cn">shangding.gu@sjtu.edu.cn</a> ·
<a href="https://github.com/SAIL-Research-Lab" target="_blank">GitHub</a> ·
<a href="https://www.youtube.com/@ai-agent-research" target="_blank">YouTube</a> ·
<a href="https://scholar.google.com/citations?user=E1GCDXUAAAAJ&hl=en" target="_blank">Google Scholar</a>
</div>
</footer>
<script>
document.querySelectorAll('.nav-item a').forEach(function(a){
a.addEventListener('click', function(){ document.body.classList.remove('nav-open'); });
});
</script>
</body>
</html>