-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathresearch.html
More file actions
692 lines (643 loc) · 52.5 KB
/
Copy pathresearch.html
File metadata and controls
692 lines (643 loc) · 52.5 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
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title>Research — LeCAR Lab</title>
<link rel="stylesheet" href="styles.css">
<link rel="icon" href="assets/icon.jpg">
<link rel="preconnect" href="https://fonts.googleapis.com">
<link rel="preconnect" href="https://fonts.gstatic.com" crossorigin>
<link href="https://fonts.googleapis.com/css2?family=Source+Serif+4:ital,opsz,wght@0,8..60,400..700;1,8..60,400..700&family=Source+Sans+3:ital,wght@0,400..700;1,400..700&display=swap" rel="stylesheet">
<link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/font-awesome/5.15.3/css/all.min.css">
<script>
// Apply saved theme before paint to avoid a flash of the wrong theme.
(function () {
try {
var t = localStorage.getItem('theme');
if (t === 'dark' || t === 'light') document.documentElement.setAttribute('data-theme', t);
} catch (e) {}
})();
</script>
</head>
<body>
<header class="site-header">
<div class="container nav-row">
<a class="brand" href="index.html"><img src="assets/logo.png" alt="LeCAR Lab — Learning and Control for Agile Robotics Lab"></a>
<input type="checkbox" id="nav-toggle" class="nav-toggle" aria-label="Toggle navigation">
<label for="nav-toggle" class="nav-toggle-label" aria-hidden="true"><i class="fas fa-bars"></i></label>
<nav class="site-nav">
<a href="people.html">People</a>
<a href="publications.html">Publications</a>
<a href="research.html" class="active">Research Topics</a>
<a href="join.html">Join</a>
</nav>
<button type="button" class="theme-toggle" aria-label="Toggle dark mode"><i class="fas fa-moon"></i></button>
</div>
</header>
<main class="container">
<section>
<h2>Research Talks</h2>
<ul class="talks">
<li>2026 May: Guest Lecture for Stanford CS224R "Sim2Real Robot Learning: A Holistic Overview" [<a href="https://drive.google.com/file/d/1buggW-9NeNzysDoT5VamK_rfnd0Q84lw/view?usp=sharing" target="_blank"><i class="far fa-file"></i> slides</a>]</li>
<li>2026 Jan: UCSD CRI Seminar Talk "Humanoid Learning from Human Data with Physics Grounding" [<a href="https://drive.google.com/file/d/1UZlqPraCNpKL1zSngevETnWxz_tv2gnf/view?usp=sharing" target="_blank"><i class="far fa-file"></i> slides</a>]</li>
<li>2025 Jun: CVPR Robotics 101 Tutorial "From Sim2Real 1.0 to 4.0 for Humanoid Whole-Body Control and Loco-Manipulation" (50 mins) [<a href="https://youtu.be/elY7TTiGUZ8?si=uXhFnRy5nkUepkjB" target="_blank"><i class="fas fa-globe"></i> recording</a>] [<a href="https://drive.google.com/file/d/1h5MxNHFo0e0TCRn9exdrHrPH8TcpjbWy/view?usp=sharing" target="_blank"><i class="far fa-file"></i> slides</a>]</li>
<li>2024 Oct: CMU RI Seminar Talk "Building Generalist Robots with Agility via Learning and Control: Humanoids and Beyond" (one hour) [<a href="https://youtu.be/Uym3Tr6t5TM?si=dJWQgO-mxbUlIwoQ" target="_blank"><i class="fas fa-globe"></i> recording</a>] [<a href="https://youtu.be/pVTG7809RTg?si=6BnsMqveGO_KIrdn" target="_blank"><i class="fas fa-globe"></i> 2025 April version at ETH Zürich</a>] [<a href="https://drive.google.com/file/d/1iyMGxjK5ID-j-ga3B7sUsqN3erle1ibY/view?usp=sharing" target="_blank"><i class="far fa-file"></i> slides</a>]</li>
<li>2024 Sep: Georgia Tech IRIM Seminar Talk "Unifying Semantic and Physical Intelligence for Generalist Humanoid Robots" (one hour) [<a href="https://mediaspace.gatech.edu/media/1_tckra9zz" target="_blank"><i class="fas fa-globe"></i> recording</a>]</li>
<!-- <li>2023 Sep: New faculty lighting talk at CMU SCS (5 mins) [<a href="https://youtu.be/1MH-R6_UALw?si=isH5XlHQzg1719jy" target="_blank"><i class="fas fa-globe"></i> recording</a>]</li> -->
</ul>
</section>
<section class="topic">
<h2>Humanoid Whole-Body Control and Behavior Foundation Model</h2>
<details>
<summary>Selected papers in this topic</summary>
<article class="pub">
<div class="pub-media">
<img src="assets/OmniH2O.gif" alt="OmniH2O">
</div>
<div>
<p class="pub-title">OmniH2O: Universal and Dexterous Human-to-Humanoid Whole-Body Teleoperation and Learning</p>
<p class="pub-authors">Tairan He<sup>*</sup>, Zhengyi Luo<sup>*</sup>, Xialin He<sup>*</sup>, Wenli Xiao, Chong Zhang, Weinan Zhang, Kris Kitani, Changliu Liu, Guanya Shi</p>
<p class="pub-venue"><em>Conference on Robot Learning (CoRL)</em>, 2024</p>
<p class="pub-links">
<a href="https://arxiv.org/abs/2406.08858" target="_blank"><i class="far fa-file"></i> paper</a>
<a href="https://omni.human2humanoid.com/" target="_blank"><i class="fas fa-globe"></i> website</a>
<a href="https://cmu.box.com/s/kmayzq5ax2rxvwn97s0hzz0aq5vws9io" target="_blank"><i class="fas fa-database"></i> dataset</a>
<a href="https://github.com/LeCAR-Lab/human2humanoid" target="_blank"><i class="fas fa-code"></i> code</a>
<a href="https://spectrum.ieee.org/video-friday-drone-vs-flying-canoe" target="_blank"><i class="fas fa-newspaper"></i> IEEE Spectrum</a>
</p>
<p class="tldr"><i class="fas fa-comment-dots"></i> TL;DR: OmniH2O provides a universal whole-body humanoid control interface that enables diverse teleoperation and autonomy methods.</p>
</div>
</article>
<article class="pub">
<div class="pub-media">
<img src="publications/2025_BFM.gif" alt="BFM-Zero">
</div>
<div>
<p class="pub-title">BFM-Zero: A Promptable Behavioral Foundation Model for Humanoid Control Using Unsupervised Reinforcement Learning</p>
<p class="pub-authors">Yitang Li<sup>*</sup>, Zhengyi Luo<sup>*</sup>, Tonghe Zhang, Cunxi Dai, Anssi Kanervisto, Andrea Tirinzoni, Haoyang Weng, Kris Kitani, Mateusz Guzek, Ahmed Touati, Alessandro Lazaric, Matteo Pirotta<sup>†</sup>, Guanya Shi<sup>†</sup></p>
<p class="pub-venue"><em>International Conference on Learning Representations (ICLR)</em>, 2026</p>
<p class="pub-links">
<a href="https://arxiv.org/abs/2511.04131" target="_blank"><i class="far fa-file"></i> paper</a>
<a href="https://lecar-lab.github.io/BFM-Zero/" target="_blank"><i class="fas fa-globe"></i> website</a>
<a href="https://github.com/LeCAR-Lab/BFM-Zero" target="_blank"><i class="fas fa-code"></i> code</a>
</p>
<p class="tldr"><i class="fas fa-comment-dots"></i> TL;DR: BFM-Zero enables zero-shot goal reaching, tracking, and reward optimization (any reward at test time) from one policy.</p>
</div>
</article>
<article class="pub">
<div class="pub-media">
<img src="assets/ASAP.gif" alt="ASAP">
</div>
<div>
<p class="pub-title">ASAP: Aligning Simulation and Real-World Physics for Learning Agile Humanoid Whole-Body Skills</p>
<p class="pub-authors">Tairan He<sup>*</sup>, Jiawei Gao<sup>*</sup>, Wenli Xiao<sup>*</sup>, Yuanhang Zhang<sup>*</sup>, Zi Wang, Jiashun Wang, Zhengyi Luo, Guanqi He, Nikhil Sobanbabu, Chaoyi Pan, Zeji Yi, Guannan Qu, Kris Kitani, Jessica Hodgins, Linxi "Jim" Fan, Yuke Zhu, Changliu Liu, Guanya Shi</p>
<p class="pub-venue"><em>Robotics: Science and Systems (RSS)</em>, 2025</p>
<p class="pub-links">
<a href="https://arxiv.org/abs/2502.01143" target="_blank"><i class="far fa-file"></i> paper</a>
<a href="https://agile.human2humanoid.com/" target="_blank"><i class="fas fa-globe"></i> website</a>
<a href="https://github.com/LeCAR-Lab/ASAP" target="_blank"><i class="fas fa-code"></i> code</a>
</p>
<p class="tldr"><i class="fas fa-comment-dots"></i> TL;DR: ASAP learns agile whole-body humanoid motions via learning a residual action model from the real world to align sim and real physics.</p>
</div>
</article>
<article class="pub">
<div class="pub-media">
<img src="publications/2025_TWIST2.gif" alt="TWIST2">
</div>
<div>
<p class="pub-title">TWIST2: Scalable, Portable, and Holistic Humanoid Data Collection System</p>
<p class="pub-authors">Yanjie Ze, Siheng Zhao, Weizhuo Wang, Angjoo Kanazawa<sup>†</sup>, Rocky Duan<sup>†</sup>, Pieter Abbeel<sup>†</sup>, Guanya Shi<sup>†</sup>, Jiajun Wu<sup>†</sup>, C. Karen Liu<sup>†</sup></p>
<p class="pub-venue"><em>International Conference on Robotics and Automation (ICRA)</em>, 2026</p>
<p class="pub-links">
<a href="https://arxiv.org/abs/2511.02832" target="_blank"><i class="far fa-file"></i> paper</a>
<a href="https://yanjieze.com/TWIST2/" target="_blank"><i class="fas fa-globe"></i> website</a>
<a href="https://github.com/amazon-far/TWIST2" target="_blank"><i class="fas fa-code"></i> code</a>
</p>
<p class="tldr"><i class="fas fa-comment-dots"></i> TL;DR: TWIST2 is a portable whole-body humanoid teleoperation system that enables scalable data collection.</p>
</div>
</article>
</details>
</section>
<section class="topic">
<h2>Humanoid Loco-Manipulation and Scene Interaction</h2>
<details>
<summary>Selected papers in this topic</summary>
<article class="pub">
<div class="pub-media">
<img src="publications/2025_OmniRetarget.gif" alt="OmniRetarget">
</div>
<div>
<p class="pub-title">OmniRetarget: Interaction-Preserving Data Generation for Humanoid Whole-Body Loco-Manipulation and Scene Interaction</p>
<p class="pub-authors">Lujie Yang<sup>*</sup>, Xiaoyu Huang<sup>*</sup>, Zhen Wu<sup>*</sup>, Angjoo Kanazawa<sup>†</sup>, Pieter Abbeel<sup>†</sup>, Carmelo Sferrazza<sup>†</sup>, C. Karen Liu<sup>†</sup>, Rocky Duan<sup>†</sup>, Guanya Shi<sup>†</sup></p>
<p class="pub-venue"><em>International Conference on Robotics and Automation (ICRA)</em>, 2026</p>
<p class="award">(Best Conference Paper Award)</p>
<p class="award">(Best Paper Award on Robot Manipulation and Locomotion)</p>
<p class="pub-links">
<a href="https://arxiv.org/abs/2509.26633" target="_blank"><i class="far fa-file"></i> paper</a>
<a href="https://omniretarget.github.io/" target="_blank"><i class="fas fa-globe"></i> website</a>
<a href="https://huggingface.co/datasets/omniretarget/OmniRetarget_Dataset" target="_blank"><i class="fas fa-database"></i> dataset</a>
</p>
<p class="tldr"><i class="fas fa-comment-dots"></i> TL;DR: High-quality interaction-preserving motion reference generation that enables agile whole-body skills with minimal RL tracking.</p>
</div>
</article>
<article class="pub">
<div class="pub-media">
<img src="publications/2026_PHP.gif" alt="Perceptive Humanoid Parkour">
</div>
<div>
<p class="pub-title">Perceptive Humanoid Parkour: Chaining Dynamic Human Skills via Motion Matching</p>
<p class="pub-authors">Zhen Wu<sup>*</sup>, Xiaoyu Huang<sup>*</sup>, Lujie Yang<sup>*</sup>, Yuanhang Zhang, Xi Chen, Pieter Abbeel<sup>†</sup>, Rocky Duan<sup>†</sup>, Angjoo Kanazawa<sup>†</sup>, Carmelo Sferrazza<sup>†</sup>, Guanya Shi<sup>†</sup>, C. Karen Liu<sup>†</sup></p>
<p class="pub-venue"><em>Robotics: Science and Systems (RSS)</em>, 2026</p>
<p class="pub-links">
<a href="https://arxiv.org/abs/2602.15827" target="_blank"><i class="far fa-file"></i> paper</a>
<a href="https://php-parkour.github.io/" target="_blank"><i class="fas fa-globe"></i> website</a>
</p>
<p class="tldr"><i class="fas fa-comment-dots"></i> TL;DR: PHP enables humanoid robots to autonomously perform long-horizon, vision-based parkour across challenging obstacle courses.</p>
</div>
</article>
<article class="pub">
<div class="pub-media">
<img src="assets/FALCON.gif" alt="FALCON">
</div>
<div>
<p class="pub-title">FALCON: Learning Force-Adaptive Humanoid Loco-Manipulation</p>
<p class="pub-authors">Yuanhang Zhang, Yifu Yuan, Prajwal Gurunath, Tairan He, Shayegan Omidshafiei, Ali-akbar Agha-mohammadi, Marcell Vazquez-Chanlatte, Liam Pedersen, Guanya Shi</p>
<p class="pub-venue"><em>Learning for Dynamics and Control Conference (L4DC)</em>, 2026</p>
<p class="award">(Oral Presentation)</p>
<p class="pub-links">
<a href="https://arxiv.org/abs/2505.06776" target="_blank"><i class="far fa-file"></i> paper</a>
<a href="https://lecar-lab.github.io/falcon-humanoid/" target="_blank"><i class="fas fa-globe"></i> website</a>
<a href="https://github.com/LeCAR-Lab/FALCON/" target="_blank"><i class="fas fa-code"></i> code</a>
</p>
<p class="tldr"><i class="fas fa-comment-dots"></i> TL;DR: FALCON enables various heavy-duty humanoid loco-manipulation tasks via a new dual-agent force-adaptive RL framework.</p>
</div>
</article>
<article class="pub">
<div class="pub-media">
<img src="publications/2025_VIRAL.gif" alt="VIRAL">
</div>
<div>
<p class="pub-title">VIRAL: Visual Sim-to-Real at Scale for Humanoid Loco-Manipulation</p>
<p class="pub-authors">Tairan He<sup>*</sup>, Zi Wang<sup>*</sup>, Haoru Xue<sup>*</sup>, Qingwei Ben<sup>*</sup>, Zhengyi Luo, Wenli Xiao, Ye Yuan, Xingye Da, Fernando Castañeda, Shankar Sastry, Changliu Liu, Guanya Shi, Linxi Fan<sup>†</sup>, Yuke Zhu<sup>†</sup></p>
<p class="pub-venue"><em>IEEE / CVF Computer Vision and Pattern Recognition Conference (CVPR)</em>, 2026</p>
<p class="pub-links">
<a href="https://arxiv.org/abs/2511.15200" target="_blank"><i class="far fa-file"></i> paper</a>
<a href="https://viral-humanoid.github.io/" target="_blank"><i class="fas fa-globe"></i> website</a>
</p>
<p class="tldr"><i class="fas fa-comment-dots"></i> TL;DR: VIRAL investigates the scaling law of visual sim-to-real and finds a recipe to achieve zero-shot, robust, and continuous real-world deployment.</p>
</div>
</article>
</details>
</section>
<section class="topic">
<h2>Contextual Locomotion</h2>
<details>
<summary>Selected papers in this topic</summary>
<article class="pub">
<div class="pub-media">
<img src="publications/2026_RPL.gif" alt="RPL">
</div>
<div>
<p class="pub-title">RPL: Learning Robust Humanoid Perceptive Locomotion on Challenging Terrains</p>
<p class="pub-authors">Yuanhang Zhang, Younggyo Seo, Juyue Chen, Yifu Yuan, Koushil Sreenath, Pieter Abbeel<sup>†</sup>, Carmelo Sferrazza<sup>†</sup>, Karen Liu<sup>†</sup>, Rocky Duan<sup>†</sup>, Guanya Shi<sup>†</sup></p>
<p class="pub-venue"><em>arXiv preprint</em></p>
<p class="pub-links">
<a href="https://arxiv.org/abs/2602.03002" target="_blank"><i class="far fa-file"></i> paper</a>
<a href="https://rpl-humanoid.github.io/" target="_blank"><i class="fas fa-globe"></i> website</a>
</p>
<p class="tldr"><i class="fas fa-comment-dots"></i> TL;DR: A single policy trained by RPL enables multi-directional robust humanoid locomotion over various challenging terrains.</p>
</div>
</article>
<article class="pub">
<div class="pub-media">
<img src="assets/abs.gif" alt="Agile But Safe">
</div>
<div>
<p class="pub-title">Agile But Safe: Learning Collision-Free High-Speed Legged Locomotion</p>
<p class="pub-authors">Tairan He<sup>*</sup>, Chong Zhang<sup>*</sup>, Wenli Xiao, Guanqi He, Changliu Liu, Guanya Shi</p>
<p class="pub-venue"><em>Robotics: Science and Systems (RSS)</em>, 2024</p>
<p class="award">(Outstanding Student Paper Award Finalist)</p>
<p class="pub-links">
<a href="https://arxiv.org/abs/2401.17583" target="_blank"><i class="far fa-file"></i> paper</a>
<a href="https://agile-but-safe.github.io/" target="_blank"><i class="fas fa-globe"></i> website</a>
<a href="https://github.com/LeCAR-Lab/ABS" target="_blank"><i class="fas fa-code"></i> code</a>
<a href="https://spectrum.ieee.org/video-friday-agile-but-safe" target="_blank"><i class="fas fa-newspaper"></i> IEEE Spectrum</a>
<a href="https://www.ri.cmu.edu/collision-free-high-speed-robots/" target="_blank"><i class="fas fa-newspaper"></i> CMU News</a>
</p>
<p class="tldr"><i class="fas fa-comment-dots"></i> TL;DR: ABS enables fully onboard, agile (>3m/s), and collision-free locomotion for quadrupedal robots in cluttered environments.</p>
</div>
</article>
<article class="pub">
<div class="pub-media">
<video src="publications/2026_LadderMan.mp4" controls autoplay loop muted playsinline></video>
</div>
<div>
<p class="pub-title">LadderMan: Learning Humanoid Perceptive Ladder Climbing</p>
<p class="pub-authors">Siheng Zhao, Yuanhang Zhang, Ziqi Lu, Pieter Abbeel, Rocky Duan, Koushil Sreenath, Yue Wang, C. Karen Liu<sup>†</sup>, Guanya Shi<sup>†</sup></p>
<p class="pub-venue"><em>arXiv preprint</em></p>
<p class="pub-links">
<a href="https://arxiv.org/abs/2606.05873" target="_blank"><i class="far fa-file"></i> paper</a>
<a href="https://ladderman-robot.github.io/" target="_blank"><i class="fas fa-globe"></i> website</a>
<a href="https://github.com/amazon-far/LadderMan" target="_blank"><i class="fas fa-code"></i> code</a>
</p>
<p class="tldr"><i class="fas fa-comment-dots"></i> TL;DR: LadderMan enables humanoids to climb diverse ladders zero-shot and perform stable manipulation while balancing on the ladder.</p>
</div>
</article>
</details>
</section>
<section class="topic">
<h2>Learning Dexterous Skills from Human Data with Physics Grounding</h2>
<details>
<summary>Selected papers in this topic</summary>
<article class="pub">
<div class="pub-media">
<img src="publications/2025_SPIDER.gif" alt="SPIDER">
</div>
<div>
<p class="pub-title">SPIDER: Scalable Physics-Informed Dexterous Retargeting</p>
<p class="pub-authors">Chaoyi Pan, Changhao Wang, Haozhi Qi, Zixi Liu, Homanga Bharadhwaj, Akash Sharma, Tingfan Wu<sup>†</sup>, Guanya Shi<sup>†</sup>, Jitendra Malik<sup>†</sup>, Francois Hogan<sup>†</sup></p>
<p class="pub-venue"><em>International Conference on Intelligent Robots and Systems (IROS)</em>, 2026</p>
<p class="pub-links">
<a href="https://arxiv.org/abs/2511.09484" target="_blank"><i class="far fa-file"></i> paper</a>
<a href="https://jc-bao.github.io/spider-project/" target="_blank"><i class="fas fa-globe"></i> website</a>
<a href="https://github.com/jc-bao/spider" target="_blank"><i class="fas fa-code"></i> code</a>
</p>
<p class="tldr"><i class="fas fa-comment-dots"></i> TL;DR: A dynamically feasible, cross-embodiment retargeting framework for both humanoid and dexterous hand. Human → physics → real at scale.</p>
</div>
</article>
<article class="pub">
<div class="pub-media">
<video src="publications/2026_LUCID.mp4" controls autoplay loop muted playsinline></video>
</div>
<div>
<p class="pub-title">LUCID: Learning Embodiment-Agnostic Intent Models from Unstructured Human Videos for Scalable Dexterous Robot Skill Acquisition</p>
<p class="pub-authors">Harsh Gupta, Guanya Shi<sup>†</sup>, Wenzhen Yuan<sup>†</sup></p>
<p class="pub-venue"><em>arXiv preprint</em></p>
<p class="pub-links">
<a href="https://arxiv.org/abs/2606.11628" target="_blank"><i class="far fa-file"></i> paper</a>
<a href="https://lucid-robot.github.io/" target="_blank"><i class="fas fa-globe"></i> website</a>
</p>
<p class="tldr"><i class="fas fa-comment-dots"></i> TL;DR: LUCID separates human-video intent from sim-trained control, enabling scalable real-world dexterous manipulation across tasks and embodiments w/o any real robot data.</p>
</div>
</article>
<article class="pub">
<div class="pub-media">
<img src="publications/2025_HDMI.gif" alt="HDMI">
</div>
<div>
<p class="pub-title">HDMI: Learning Interactive Humanoid Whole-Body Control from Human Videos</p>
<p class="pub-authors">Haoyang Weng, Yitang Li, Nikhil Sobanbabu, Zihan Wang, Zhengyi Luo, Tairan He, Deva Ramanan, Guanya Shi</p>
<p class="pub-venue"><em>arXiv preprint</em></p>
<p class="pub-links">
<a href="https://arxiv.org/abs/2509.16757" target="_blank"><i class="far fa-file"></i> paper</a>
<a href="https://hdmi-humanoid.github.io/#/" target="_blank"><i class="fas fa-globe"></i> website</a>
<a href="https://github.com/LeCAR-Lab/HDMI" target="_blank"><i class="fas fa-code"></i> code</a>
</p>
<p class="tldr"><i class="fas fa-comment-dots"></i> TL;DR: From human videos, HDMI learns robust humanoid loco-manipulation skills (e.g., opening a door continuously for 67 times).</p>
</div>
</article>
</details>
</section>
<section class="topic">
<h2>Policy Adaptation and Self-Improvement in the Real World</h2>
<details>
<summary>Selected papers in this topic</summary>
<article class="pub">
<div class="pub-media">
<img src="publications/2022_neural_fly.gif" alt="Neural-Fly">
</div>
<div>
<p class="pub-title">Neural-Fly Enables Rapid Learning for Agile Flight in Strong Winds</p>
<p class="pub-authors">Michael O'Connell<sup>*</sup>, Guanya Shi<sup>*</sup>, Xichen Shi, Kamyar Azizzadenesheli, Animashree Anandkumar, Yisong Yue, Soon-Jo Chung</p>
<p class="pub-venue"><em>Science Robotics</em></p>
<p class="pub-links">
<a href="https://arxiv.org/abs/2205.06908" target="_blank"><i class="far fa-file"></i> paper</a>
<a href="https://youtu.be/TuF9teCZX0U" target="_blank"><i class="fas fa-video"></i> video</a>
<a href="https://www.caltech.edu/about/news/rapid-adaptation-of-deep-learning-teaches-drones-to-survive-any-weather" target="_blank"><i class="fas fa-newspaper"></i> Caltech news</a>
<a href="https://youtu.be/R1S5BnKgJxs" target="_blank"><i class="fas fa-newspaper"></i> Reuters</a>
<a href="https://www.cnn.com/videos/business/2022/05/31/caltech-neural-fly-drones-in-strong-wind-orig-ht.cnn-business/video/playlists/business-tech/" target="_blank"><i class="fas fa-newspaper"></i> CNN</a>
<a href="https://github.com/aerorobotics/neural-fly" target="_blank"><i class="fas fa-code"></i> code</a>
</p>
<p class="tldr"><i class="fas fa-comment-dots"></i> TL;DR: Neural-Fly uses adaptive control to online fine-tune a meta-pretrained DNN representation, enabling rapid adaptation in strong winds.</p>
</div>
</article>
<article class="pub">
<div class="pub-media">
<img src="publications/2024_AnyCar.gif" alt="AnyCar">
</div>
<div>
<p class="pub-title">AnyCar to Anywhere: Learning Universal Dynamics Model for Agile and Adaptive Mobility</p>
<p class="pub-authors">Wenli Xiao<sup>*</sup>, Haoru Xue<sup>*</sup>, Tony Tao, Dvij Kalaria, John M. Dolan, Guanya Shi</p>
<p class="pub-venue"><em>International Conference on Robotics and Automation (ICRA)</em>, 2025</p>
<p class="pub-links">
<a href="https://arxiv.org/abs/2409.15783" target="_blank"><i class="far fa-file"></i> paper</a>
<a href="https://lecar-lab.github.io/anycar/" target="_blank"><i class="fas fa-globe"></i> website</a>
<a href="https://github.com/LeCAR-Lab/anycar" target="_blank"><i class="fas fa-code"></i> code</a>
<a href="https://spectrum.ieee.org/video-friday-mobile-robot-upgrades" target="_blank"><i class="fas fa-newspaper"></i> IEEE Spectrum</a>
</p>
<p class="tldr"><i class="fas fa-comment-dots"></i> TL;DR: AnyCar is a transformer-based dynamics model that can adapt to various vehicles, environments, state estimators, and tasks.</p>
</div>
</article>
<article class="pub">
<div class="pub-media">
<img src="publications/2025_PLD.gif" alt="PLD">
</div>
<div>
<p class="pub-title">Self-Improving Vision-Language-Action Models with Data Generation via Residual RL</p>
<p class="pub-authors">Wenli Xiao<sup>*</sup>, Haotian Lin<sup>*</sup>, Andy Peng, Haoru Xue, Tairan He, Yuqi Xie, Fengyuan Hu, Jimmy Wu, Zhengyi Luo, Linxi "Jim" Fan, Guanya Shi, Yuke Zhu</p>
<p class="pub-venue"><em>International Conference on Learning Representations (ICLR)</em>, 2026</p>
<p class="pub-links">
<a href="https://arxiv.org/abs/2511.00091" target="_blank"><i class="far fa-file"></i> paper</a>
<a href="https://wenlixiao.com/self-improve-VLA-PLD" target="_blank"><i class="fas fa-globe"></i> website</a>
</p>
<p class="tldr"><i class="fas fa-comment-dots"></i> TL;DR: Probe, Learn, Distill (PLD): On-policy probing from a base VLA model + off-policy residual RL + distillation for VLA post-training.</p>
</div>
</article>
<article class="pub">
<div class="pub-media">
<video src="publications/2026_ENPIRE.mp4" controls autoplay loop muted playsinline></video>
</div>
<div>
<p class="pub-title">ENPIRE: Agentic Robot Policy Self-Improvement in the Real World</p>
<p class="pub-authors">Wenli Xiao<sup>*</sup>, Jia Xie<sup>*</sup>, Tonghe Zhang<sup>*</sup>, Haotian Lin<sup>*</sup>, Letian "Max" Fu, Haoru Xue, Jalen Lu, Yi Yang, Cunxi Dai, Zi Wang, Jimmy Wu, Guanzhi Wang, S. Shankar Sastry, Ken Goldberg, Linxi "Jim" Fan<sup>†</sup>, Yuke Zhu<sup>†</sup>, Guanya Shi<sup>†</sup></p>
<p class="pub-venue"><em>arXiv preprint</em></p>
<p class="pub-links">
<a href="https://research.nvidia.com/labs/gear/enpire/" target="_blank"><i class="fas fa-globe"></i> website</a>
</p>
<p class="tldr"><i class="fas fa-comment-dots"></i> TL;DR: ENPIRE gives tool-calling coding agents a real-world feedback loop, enabling autonomous policy self-improvement to 99% success on dexterous manipulation tasks.</p>
</div>
</article>
</details>
</section>
<section class="topic">
<h2>Robot Learning "Science"</h2>
<details>
<summary>Selected papers in this topic</summary>
<article class="pub">
<div class="pub-media">
<img src="publications/2025_MIP.jpg" alt="Much Ado About Noising">
</div>
<div>
<p class="pub-title">Much Ado About Noising: Dispelling the Myths of Generative Robotic Control</p>
<p class="pub-authors">Chaoyi Pan, Giri Anantharaman, Nai-Chieh Huang, Claire Jin, Daniel Pfrommer, Chenyang Yuan, Frank Permenter, Guannan Qu<sup>†</sup>, Nicholas Boffi<sup>†</sup>, Guanya Shi<sup>†</sup>, Max Simchowitz<sup>†</sup></p>
<p class="pub-venue"><em>International Conference on Learning Representations (ICLR)</em>, 2026</p>
<p class="pub-links">
<a href="https://arxiv.org/abs/2512.01809" target="_blank"><i class="far fa-file"></i> paper</a>
<a href="https://simchowitzlabpublic.github.io/much-ado-about-noising-project/" target="_blank"><i class="fas fa-globe"></i> website</a>
<a href="https://github.com/simchowitzlabpublic/much-ado-about-noising" target="_blank"><i class="fas fa-code"></i> code</a>
</p>
<p class="tldr"><i class="fas fa-comment-dots"></i> TL;DR: In most benchmarks, the success of generative policies is NOT from its distributional-learning formulation.</p>
</div>
</article>
</details>
</section>
<section class="topic">
<h2>Agentic Robot Learning</h2>
<details>
<summary>Selected papers in this topic</summary>
<article class="pub">
<div class="pub-media">
<video src="publications/2026_ENPIRE.mp4" controls autoplay loop muted playsinline></video>
</div>
<div>
<p class="pub-title">ENPIRE: Agentic Robot Policy Self-Improvement in the Real World</p>
<p class="pub-authors">Wenli Xiao<sup>*</sup>, Jia Xie<sup>*</sup>, Tonghe Zhang<sup>*</sup>, Haotian Lin<sup>*</sup>, Letian "Max" Fu, Haoru Xue, Jalen Lu, Yi Yang, Cunxi Dai, Zi Wang, Jimmy Wu, Guanzhi Wang, S. Shankar Sastry, Ken Goldberg, Linxi "Jim" Fan<sup>†</sup>, Yuke Zhu<sup>†</sup>, Guanya Shi<sup>†</sup></p>
<p class="pub-venue"><em>arXiv preprint</em></p>
<p class="pub-links">
<a href="https://research.nvidia.com/labs/gear/enpire/" target="_blank"><i class="fas fa-globe"></i> website</a>
</p>
<p class="tldr"><i class="fas fa-comment-dots"></i> TL;DR: ENPIRE gives tool-calling coding agents a real-world feedback loop, enabling autonomous policy self-improvement to 99% success on dexterous manipulation tasks.</p>
</div>
</article>
<article class="pub">
<div class="pub-media">
<img src="publications/2026_CapX.gif" alt="CaP-X">
</div>
<div>
<p class="pub-title">CaP-X: A Framework for Benchmarking and Improving Coding Agents for Robot Manipulation</p>
<p class="pub-authors">Max Fu<sup>*</sup>, Justin Yu<sup>*</sup>, Karim El-Refai<sup>*</sup>, Ethan Kou<sup>*</sup>, Haoru Xue<sup>*</sup>, Huang Huang, Wenli Xiao, Guanzhi Wang, Fei-Fei Li, Guanya Shi, Jiajun Wu, Shankar Sastry, Yuke Zhu, Ken Goldberg<sup>†</sup>, Linxi "Jim" Fan<sup>†</sup></p>
<p class="pub-venue"><em>International Conference on Machine Learning (ICML)</em>, 2026</p>
<p class="pub-links">
<a href="https://arxiv.org/abs/2603.22435" target="_blank"><i class="far fa-file"></i> paper</a>
<a href="https://capgym.github.io/" target="_blank"><i class="fas fa-globe"></i> website</a>
<a href="https://github.com/capgym/cap-x" target="_blank"><i class="fas fa-code"></i> code</a>
</p>
<p class="tldr"><i class="fas fa-comment-dots"></i> TL;DR: CaP-X benchmarks and improves embodied coding agents, enabling LMs to write robot-control code that generalizes zero-shot.</p>
</div>
</article>
</details>
</section>
<section class="topic">
<h2>Sampling-Based Optimal Control, MBRL, and World Model</h2>
<details>
<summary>Selected papers in this topic</summary>
<article class="pub">
<div class="pub-media">
<img src="publications/2024_DIAL-MPC.gif" alt="DIAL-MPC">
</div>
<div>
<p class="pub-title">Full-Order Sampling-Based MPC for Torque-Level Locomotion Control via Diffusion-Style Annealing</p>
<p class="pub-authors">Haoru Xue<sup>*</sup>, Chaoyi Pan<sup>*</sup>, Zeji Yi, Guannan Qu, Guanya Shi</p>
<p class="pub-venue"><em>International Conference on Robotics and Automation (ICRA)</em>, 2025</p>
<p class="award">(Best Paper Award Finalist)</p>
<p class="pub-links">
<a href="https://arxiv.org/abs/2409.15610" target="_blank"><i class="far fa-file"></i> paper</a>
<a href="https://lecar-lab.github.io/dial-mpc/" target="_blank"><i class="fas fa-globe"></i> website</a>
<a href="https://github.com/LeCAR-Lab/dial-mpc" target="_blank"><i class="fas fa-code"></i> code</a>
</p>
<p class="tldr"><i class="fas fa-comment-dots"></i> TL;DR: DIAL-MPC is the first training-free method achieving real-time whole-body torque control using full-order dynamics.</p>
</div>
</article>
<article class="pub">
<div class="pub-media">
<img src="publications/2025_TDMPC_Square.gif" alt="TD-M(PC)²">
</div>
<div>
<p class="pub-title">TD-M(PC)<sup>2</sup>: Improving Temporal Difference MPC Through Policy Constraint</p>
<p class="pub-authors">Haotian Lin, Pengcheng Wang, Jeff Schneider, Guanya Shi</p>
<p class="pub-venue"><em>Learning for Dynamics and Control Conference (L4DC)</em>, 2026</p>
<p class="pub-links">
<a href="https://arxiv.org/abs/2502.03550" target="_blank"><i class="far fa-file"></i> paper</a>
<a href="https://darthutopian.github.io/tdmpc_square/" target="_blank"><i class="fas fa-globe"></i> website</a>
<a href="https://github.com/DarthUtopian/tdmpc_square_public" target="_blank"><i class="fas fa-code"></i> code</a>
</p>
<p class="tldr"><i class="fas fa-comment-dots"></i> TL;DR: We observe the value overestimation issue in planner-based MBRL and propose a policy constraint solution with SOTA performance.</p>
</div>
</article>
<article class="pub">
<div class="pub-media">
<img src="publications/2024_MBD.gif" alt="Model-Based Diffusion">
</div>
<div>
<p class="pub-title">Model-Based Diffusion for Trajectory Optimization</p>
<p class="pub-authors">Chaoyi Pan<sup>*</sup>, Zeji Yi<sup>*</sup>, Guanya Shi<sup>†</sup>, Guannan Qu<sup>†</sup></p>
<p class="pub-venue"><em>Neural Information Processing Systems (NeurIPS)</em>, 2024</p>
<p class="pub-links">
<a href="https://drive.google.com/file/d/1kPjD79Cfr9spWulWNVFMRHqTE-mjbGAp/view?usp=sharing" target="_blank"><i class="far fa-file"></i> paper</a>
<a href="https://lecar-lab.github.io/mbd/" target="_blank"><i class="fas fa-globe"></i> website</a>
<a href="https://github.com/LeCAR-Lab/model-based-diffusion" target="_blank"><i class="fas fa-code"></i> code</a>
</p>
<p class="tldr"><i class="fas fa-comment-dots"></i> TL;DR: MBD is a diffusion-based traj optimization method that directly computes the score function using models without any external data.</p>
</div>
</article>
</details>
</section>
<section class="topic">
<h2>Embodied AI in the Air: General-Purpose Aerial Manipulation</h2>
<details>
<summary>Selected papers in this topic</summary>
<article class="pub">
<div class="pub-media">
<img src="publications/2024_flying_calligrapher.gif" alt="Flying Calligrapher">
</div>
<div>
<p class="pub-title">Flying Calligrapher: Contact-Aware Motion and Force Planning and Control for Aerial Manipulation</p>
<p class="pub-authors">Xiaofeng Guo<sup>*</sup>, Guanqi He<sup>*</sup>, Jiahe Xu, Mohammadreza Mousaei, Junyi Geng, Sebastian Scherer, Guanya Shi</p>
<p class="pub-venue"><em>IEEE Robotics and Automation Letters (RA-L)</em>, 2024</p>
<p class="pub-links">
<a href="https://arxiv.org/abs/2407.05587" target="_blank"><i class="far fa-file"></i> paper</a>
<a href="https://xiaofeng-guo.github.io/flying-calligrapher/" target="_blank"><i class="fas fa-globe"></i> website</a>
<a href="https://spectrum.ieee.org/video-friday-unitree-talks-robots" target="_blank"><i class="fas fa-newspaper"></i> IEEE Spectrum</a>
</p>
<p class="tldr"><i class="fas fa-comment-dots"></i> TL;DR: Flying calligrapher enables precise hybrid motion and contact force control for an aerial manipulator in various drawing tasks.</p>
</div>
</article>
<article class="pub">
<div class="pub-media">
<img src="publications/2025_Flying_Hand.gif" alt="Flying Hand">
</div>
<div>
<p class="pub-title">Flying Hand: End-Effector-Centric Framework for Versatile Aerial Manipulation Teleoperation and Policy Learning</p>
<p class="pub-authors">Guanqi He<sup>*</sup>, Xiaofeng Guo<sup>*</sup>, Luyi Tang, Yuanhang Zhang, Mohammadreza Mousaei, Jiahe Xu, Junyi Geng, Sebastian Scherer, Guanya Shi</p>
<p class="pub-venue"><em>Robotics: Science and Systems (RSS)</em>, 2025</p>
<p class="pub-links">
<a href="https://arxiv.org/abs/2504.10334" target="_blank"><i class="far fa-file"></i> paper</a>
<a href="https://lecar-lab.github.io/flying_hand/" target="_blank"><i class="fas fa-globe"></i> website</a>
</p>
<p class="tldr"><i class="fas fa-comment-dots"></i> TL;DR: A general-purpose aerial manipulation framework with an EE-centric interface that bridges whole-body control and policy learning.</p>
</div>
</article>
<article class="pub">
<div class="pub-media">
<img src="publications/2025_EADP.gif" alt="UMI-on-Air">
</div>
<div>
<p class="pub-title">UMI-on-Air: Embodiment-Aware Guidance for Embodiment-Agnostic Visuomotor Policies</p>
<p class="pub-authors">Harsh Gupta, Xiaofeng Guo, Huy Ha, Chuer Pan, Muqing Cao, Dongjae Lee, Sebastian Scherer, Shuran Song, Guanya Shi</p>
<p class="pub-venue"><em>International Conference on Robotics and Automation (ICRA)</em>, 2026</p>
<p class="pub-links">
<a href="https://arxiv.org/abs/2510.02614" target="_blank"><i class="far fa-file"></i> paper</a>
<a href="https://umi-on-air.github.io/" target="_blank"><i class="fas fa-globe"></i> website</a>
</p>
<p class="tldr"><i class="fas fa-comment-dots"></i> TL;DR: EADP steers UMI's embodiment-agnostic diffusion policy using the gradient of the low-level controller's tracking cost for cross-embodiment.</p>
</div>
</article>
</details>
</section>
<section class="topic">
<h2>Structured RL and Control with Safety and Performance Guarantees</h2>
<details>
<summary>Selected papers in this topic</summary>
<article class="pub">
<div class="pub-media">
<img src="assets/abs.gif" alt="Agile But Safe">
</div>
<div>
<p class="pub-title">Agile But Safe: Learning Collision-Free High-Speed Legged Locomotion</p>
<p class="pub-authors">Tairan He<sup>*</sup>, Chong Zhang<sup>*</sup>, Wenli Xiao, Guanqi He, Changliu Liu, Guanya Shi</p>
<p class="pub-venue"><em>Robotics: Science and Systems (RSS)</em>, 2024</p>
<p class="award">(Outstanding Student Paper Award Finalist)</p>
<p class="pub-links">
<a href="https://arxiv.org/abs/2401.17583" target="_blank"><i class="far fa-file"></i> paper</a>
<a href="https://agile-but-safe.github.io/" target="_blank"><i class="fas fa-globe"></i> website</a>
<a href="https://github.com/LeCAR-Lab/ABS" target="_blank"><i class="fas fa-code"></i> code</a>
<a href="https://spectrum.ieee.org/video-friday-agile-but-safe" target="_blank"><i class="fas fa-newspaper"></i> IEEE Spectrum</a>
<a href="https://www.ri.cmu.edu/collision-free-high-speed-robots/" target="_blank"><i class="fas fa-newspaper"></i> CMU News</a>
</p>
<p class="tldr"><i class="fas fa-comment-dots"></i> TL;DR: ABS enables fully onboard, agile (>3m/s), and collision-free locomotion for quadrupedal robots in cluttered environments.</p>
</div>
</article>
<article class="pub">
<div class="pub-media">
<img src="publications/2024_JumpingCoD.gif" alt="Jumping CoD">
</div>
<div>
<p class="pub-title">Agile Continuous Jumping in Discontinuous Terrains</p>
<p class="pub-authors">Yuxiang Yang, Guanya Shi, Changyi Lin, Xiangyun Meng, Rosario Scalise, Mateo Guaman Castro, Wenhao Yu, Tingnan Zhang, Ding Zhao, Jie Tan, Byron Boots</p>
<p class="pub-venue"><em>International Conference on Robotics and Automation (ICRA)</em>, 2025</p>
<p class="pub-links">
<a href="https://arxiv.org/abs/2409.10923" target="_blank"><i class="far fa-file"></i> paper</a>
<a href="https://yxyang.github.io/jumping_cod/" target="_blank"><i class="fas fa-globe"></i> website</a>
<a href="https://github.com/yxyang/jumping_cod" target="_blank"><i class="fas fa-code"></i> code</a>
</p>
<p class="tldr"><i class="fas fa-comment-dots"></i> TL;DR: Continuous, agile, and autonomous quadrupedal jumping via hierarchical model-free RL and model-based control.</p>
</div>
</article>
<article class="pub">
<div class="pub-media">
<img src="publications/2019_neural_lander.gif" alt="Neural Lander">
</div>
<div>
<p class="pub-title">Neural Lander: Stable Drone Landing Control Using Learned Dynamics</p>
<p class="pub-authors">Guanya Shi<sup>*</sup>, Xichen Shi<sup>*</sup>, Michael O'Connell<sup>*</sup>, Rose Yu, Kamyar Azizzadenesheli, Animashree Anandkumar, Yisong Yue, Soon-Jo Chung</p>
<p class="pub-venue"><em>International Conference on Robotics and Automation (ICRA)</em>, 2019</p>
<p class="pub-links">
<a href="https://arxiv.org/abs/1811.08027" target="_blank"><i class="far fa-file"></i> paper</a>
<a href="https://youtu.be/FLLsG0S78ik" target="_blank"><i class="fas fa-video"></i> video</a>
<a href="https://www.caltech.edu/about/news/neural-lander-uses-ai-land-drones-smoothly" target="_blank"><i class="fas fa-newspaper"></i> Caltech homepage news</a>
<a href="https://github.com/GuanyaShi/neural_lander_sim_1d" target="_blank"><i class="fas fa-code"></i> code</a>
</p>
<p class="tldr"><i class="fas fa-comment-dots"></i> TL;DR: Spectrally normalized deep learning and nonlinear control enable provably stable agile drone landing.</p>
</div>
</article>
</details>
</section>
<section class="topic">
<h2>Robot Learning Infrastructure</h2>
<details>
<summary>Selected papers in this topic</summary>
<article class="pub">
<div class="pub-media">
<video src="publications/2026_ABC.mp4" controls autoplay loop muted playsinline></video>
</div>
<div>
<p class="pub-title">Scalable Behavior Cloning with Open Data, Training, and Evaluation</p>
<p class="pub-authors">Arthur Allshire<sup>*</sup>, Himanshu Gaurav Singh<sup>*</sup>, Ritvik Singh<sup>*</sup>, Adam Rashid<sup>*</sup>, Hongsuk Choi<sup>*</sup>, David McAllister<sup>*</sup>, Justin Yu, Yiyuan Chen, Huang Huang, Pieter Abbeel, Xi Chen, Rocky Duan, Phillip Isola, Jitendra Malik, Fred Shentu, Guanya Shi, Philipp Wu, Angjoo Kanazawa</p>
<p class="pub-venue"><em>arXiv preprint</em></p>
<p class="pub-links">
<a href="https://abc.bot/" target="_blank"><i class="fas fa-globe"></i> website</a>
<a href="https://github.com/amazon-far/abc" target="_blank"><i class="fas fa-code"></i> code</a>
<a href="https://huggingface.co/datasets/XDOF/ABC-130k" target="_blank"><i class="fas fa-database"></i> dataset</a>
</p>
<p class="tldr"><i class="fas fa-comment-dots"></i> TL;DR: Open data, training, and infra for robotics. We release the largest teleop dataset (>3.5K hours, >130K episodes) to date, and extensively investigate training techniques.</p>
</div>
</article>
<article class="pub">
<div class="pub-media">
<img src="publications/2026_RIO.png" alt="RIO">
</div>
<div>
<p class="pub-title">RIO: Flexible Real-Time Robot I/O for Cross-Embodiment Robot Learning</p>
<p class="pub-authors">Pablo Ortega-Kral<sup>*</sup>, Eliot Xing<sup>*</sup>, Arthur Bucker, Vernon Luk, Junseo Kim, Owen Kwon, Angchen Xie, Nikhil Sobanbabu, Yifu Yuan, Megan Lee, Deepam Ameria, Bhaswanth Ayapilla, Jaycie Bussell, Guanya Shi, Jonathan Francis, Jean Oh</p>
<p class="pub-venue"><em>Robotics: Science and Systems (RSS)</em>, 2026</p>
<p class="pub-links">
<a href="https://robot-i-o.github.io/" target="_blank"><i class="fas fa-globe"></i> website</a>
<a href="https://github.com/robot-i-o/rio" target="_blank"><i class="fas fa-code"></i> code</a>
</p>
<p class="tldr"><i class="fas fa-comment-dots"></i> TL;DR: Cross-embodiment robot learning is bottlenecked by fragmented infrastructure, not just limited data. RIO provides a flexible and real-time robot I/O framework for it.</p>
</div>
</article>
</details>
</section>
</main>
<footer class="site-footer">
<div class="social-icons">
<a href="https://youtube.com/@LeCARLab" target="_blank" aria-label="YouTube"><i class="fab fa-youtube"></i></a>
<a href="https://twitter.com/LeCARLab" target="_blank" aria-label="Twitter"><i class="fab fa-twitter"></i></a>
<a href="https://github.com/LeCAR-Lab" target="_blank" aria-label="GitHub"><i class="fab fa-github"></i></a>
</div>
</footer>
<script src="theme.js"></script>
</body>
</html>