1<!DOCTYPE html> 2<html> 3<head> 4 <meta charset="utf-8"> 5 <meta name="description" 6 content="ARGUS: Aligning Robot Geometry Under Shifting Views with Large 3D Vision Models"> 7 <meta name="keywords" content="ARGUS, VGGT"> 8 <meta name="viewport" content="width=device-width, initial-scale=1"> 9 <title>ARGUS: Aligning Robot Geometry Under Shifting Views</title> 10 11
11<script> 12
vendor: 163 bytes, lines 12-20
12window.dataLayer = window.dataLayer || []; 13 14 function gtag() { 15 dataLayer.push(arguments); 16 } 17 18 gtag('js', new Date()); 19 20 gtag('config', '
20G-PYVRSFMDRL
vendor: 7 bytes, lines 20-21
20'); 21
21</script>
21 22 23 <link href="https://fonts.googleapis.com/css?family=Google+Sans|Noto+Sans|Castoro" 24 rel="stylesheet"> 25 26 <link rel="stylesheet" href="./static/css/bulma.min.css"> 27 <link rel="stylesheet" href="./static/css/bulma-carousel.min.css"> 28 <link rel="stylesheet" href="./static/css/bulma-slider.min.css"> 29 <link rel="stylesheet" href="./static/css/fontawesome.all.min.css"> 30 <link rel="stylesheet" 31 href="https://cdn.jsdelivr.net/gh/jpswalsh/academicons@1/css/academicons.min.css"> 32 <link rel="stylesheet" href="./static/css/index.css"> 33 <link rel="icon" href="./static/images/favicon.svg"> 34 35
35<script src="https://ajax.googleapis.com/ajax/libs/jquery/3.5.1/jquery.min.js"></script>
35 36
36<script defer src="./static/js/fontawesome.all.min.js"></script>
36 37
37<script src="./static/js/bulma-carousel.min.js"></script>
37 38
38<script src="./static/js/bulma-slider.min.js"></script>
38 39
39<script src="./static/js/index.js"></script>
39 40</head> 41<body> 42 43<nav class="navbar" role="navigation" aria-label="main navigation"> 44 <div class="navbar-brand"> 45 <a role="button" class="navbar-burger" aria-label="menu" aria-expanded="false"> 46 <span aria-hidden="true"></span> 47 <span aria-hidden="true"></span> 48 <span aria-hidden="true"></span> 49 </a> 50 </div> 51 <div class="navbar-menu"> 52 <div class="navbar-start" style="flex-grow: 1; justify-content: center;"> 53 <a class="navbar-item" href="https://rsathua.github.io/"> 54 <span class="icon"> 55 <i class="fas fa-home"></i> 56 </span> 57 </a> 58 59 <div class="navbar-item has-dropdown is-hoverable"> 60 <a class="navbar-link"> 61 More Research 62 </a> 63 <div class="navbar-dropdown"> 64 <a class="navbar-item" href="https://arjung128.github.io/svm/"> 65 Servoing with Vision Models 66 </a> 67 <a class="navbar-item" href="https://arjung128.github.io/opening-cabinets-and-drawers/"> 68 MOSART 69 </a> 70 </div> 71 </div> 72 </div> 73 74 </div> 75</nav> 76 77 78<section class="hero"> 79 <div class="hero-body"> 80 <div class="container is-max-desktop"> 81 <div class="columns is-centered"> 82 <div class="column has-text-centered"> 83 <h1 class="title is-1 publication-title">ARGUS: <u>A</u>ligning <u>R</u>obot Scene <u>G</u>eometry <u>U</u>nder <u>S</u>hifting Views with Large 3D Vision Models</h1> 84 <div class="is-size-5 publication-authors"> 85 <span class="author-block"> 86 <a href="https://rsathua.github.io/">Rishik Sathua</a><sup>1</sup>,</span> 87 <span class="author-block"> 88 <a href="https://haonan16.github.io/">Haonan Chen</a><sup>2</sup>,</span> 89 <span class="author-block"> 90 <a href="https://krdc.web.illinois.edu/">Katherine Driggs-Campbell</a><sup>1</sup>, 91 </span> 92 </div> 93 94 <div class="is-size-5 publication-authors"> 95 <span class="author-block"><sup>1</sup>University of Illinois at Urbana-Champaign,</span> 96 <span class="author-block"><sup>2</sup>Harvard University</span> 97 </div> 98 99 <div class="column has-text-centered"> 100 <div class="publication-links"> 101 <!-- PDF Link. --> 102 <span class="link-block"> 103 <a href="https://arxiv.org/pdf/2608.05579" 104 class="external-link button is-normal is-rounded is-dark"> 105 <span class="icon"> 106 <i class="fas fa-file-pdf"></i> 107 </span> 108 <span>Paper</span> 109 </a> 110 </span> 111 <span class="link-block"> 112 <a href="https://arxiv.org/abs/2608.05579" 113 class="external-link button is-normal is-rounded is-dark"> 114 <span class="icon"> 115 <i class="ai ai-arxiv"></i> 116 </span> 117 <span>arXiv</span> 118 </a> 119 </span> 120 <span class="link-block"> 121 <a class="external-link button is-normal is-rounded is-dark" 122 style="pointer-events: none; opacity: 0.5; cursor: default;" 123 aria-disabled="true"> 124 <span class="icon"> 125 <i class="fab fa-github"></i> 126 </span> 127 <span>Code Coming Soon</span> 128 </a> 129 <!-- <a href="https://github.com/google/nerfies" 130 class="external-link button is-normal is-rounded is-dark"> 131 <span class="icon"> 132 <i class="fab fa-github"></i> 133 </span> 134 <span>Code</span> 135 </a> --> 136 </span> 137 <!--<span class="link-block"> 138 <a href="https://github.com/google/nerfies/releases/tag/0.1" 139 class="external-link button is-normal is-rounded is-dark">
140 <span class="icon"> 141 <i class="far fa-images"></i> 142 </span> 143 <span>Data</span> 144 </a> 145 </span>--> 146 <span class="link-block"> 147 <a href="https://www.youtube.com/watch?v=z3SqkawwJYg" 148 class="external-link button is-normal is-rounded is-dark"> 149 <span class="icon"> 150 <i class="fab fa-youtube"></i> 151 </span> 152 <span>Video</span> 153 </a> 154 </span> 155 </div> 156 157 </div> 158 </div> 159 </div> 160 </div> 161 </div> 162</section> 163 164<section class="hero teaser"> 165 <div class="container is-max-desktop"> 166 <div class="hero-body"> 167 <div class="has-text-centered"> 168 <img src="./videos/teaser.gif" width="720"> 169 </div> 170 <h2 class="subtitle has-text-centered"> 171 With ARGUS, we can efficiently train policies that maintain strong performance despite dramatic viewpoint changes. 172 </h2> 173 </div> 174 </div> 175</section> 176 177<section class="section"> 178 <div class="container is-max-desktop"> 179 <div class="columns is-centered has-text-centered"> 180 <div class="column is-four-fifths"> 181 <h2 class="title is-3">Overview</h2> 182 <div class="publication-video"> 183 <iframe src="https://www.youtube.com/embed/z3SqkawwJYg?rel=0&showinfo=0" 184 frameborder="0" allow="autoplay; encrypted-media" allowfullscreen></iframe> 185 </div> 186 </div> 187 </div> 188 </div> 189 </div> 190 </div> 191</section> 192<section class="section"> 193 <div class="container is-max-desktop"> 194 <!-- Abstract. --> 195 <div class="columns is-centered has-text-centered"> 196 <div class="column is-four-fifths"> 197 <h2 class="title is-3">Abstract</h2> 198 <img src="./images/teaser_cropped.png"> 199 <div class="content has-text-justified"> 200 <p> 201 Large-scale visuomotor policies have demonstrated impressive performance across a wide range 202 of robot manipulation tasks. However, despite this success, manipulation polices often entangle 203 scene geometry with the corresponding viewpoint, learning where objects lie in an image rather 204 than where it lies in the task space. This entanglement inherently limits the corresponding 205 policy's ability to learn from viewpoint-diverse datasets (ex. DROID, BridgeV2) and generalize 206 beyond the viewpoints captured in their training data. 207 </p> 208 <p> 209 In this work, we present ARGUS, an observation pre-processing pipeline that uses large-scale 3D 210 vision models to align image observations from arbitrary camera viewpoints into a canonical 211 viewpoint before passing it to downstream visuomotor policies. Experiments across training datasets 212 with varying levels of viewpoint diversity, from fixed multi-view camera configurations to highly 213 varied camera placements, show that our method consistently outperforms prior approaches across both 214 limited-view and view-diverse training regimes. In efficiency comparisons, ARGUS demonstrates 215 an ability to learn from view-diverse data, achieving high success rates <b>4â6Ã</b> faster than previous 216 methods by leveraging a simplified observation space. Overall, our findings indicate that, by 217 leveraging knowledge from 3D vision models, ARGUS reduces the learning burden on general 218 visuomotor policies and takes a step toward the efficient utilization of large-scale robot datasets. 219 </p> 220 </div> 221 </div> 222 </div> 223 <!--/ Abstract. --> 224 </div> 225</section> 226 227<section class="section"> 228 <div class="container is-max-widescreen"> 229 <div class="rows"> 230 <div class="rows is-centered"> 231 <div class="row is-full-width"> 232 <h2 class="title is-3 has-text-centered mb-1">Interactive Dataset Visualizer</h2> 233 <p class="d-md-block d-none has-text-centered is-size-5 mt-1 mb-4">(Showing random trajectory of 100 trajectories from put marker in cup dataset)</p> 234 235 <div class="video_grid_container"> 236 <div class="video_row"> 237 <div class="video_grid"> 238 <div class="video_grid_item"><div class="vsc-controller"></div> 239 <video id="sample-cam0" class="grid-video training" autoplay muted loop playsinline height="40%"> 240 <source id="sample-cam0-src" src="" type="video/mp4"> 241 </video> 242 </div> 243 <p>Left Camera</p> 244 </div> 245 <div class="video_grid"> 246 <div class="video_grid_item"><div class="vsc-controller"></div> 247 <video id="sample-cam1" class="grid-video training" autoplay muted loop playsinline height="40%"> 248 <source id="sample-cam1-src" src="" type="video/mp4"> 249 </video> 250 </div> 251 <p>Right Camera</p> 252 </div> 253 <div class="video_grid"> 254 <div class="video_grid_item"><div class="vsc-controller"></div> 255 <video id="sample-canonical" class="grid-video training" autoplay muted loop playsinline height="40%"> 256 <source id="sample-canonical-src" src="" type="video/mp4"> 257 </video> 258 </div> 259 <p>Canonical Rendering</p> 260 </div> 261 </div> 262 </div> 263 264 <div class="has-text-centered" style="margin-bottom: 20px;"> 265 <button id="resample-btn" class="button is-rounded is-dark" onclick="resampleTrajectory()">
266 <span class="icon"> 267 <i class="fas fa-random"></i> 268 </span> 269 <span>Resample Trajectory</span> 270 </button> 271 </div> 272 273 </div> 274 </div> 275 </div> 276 </div> <!-- rows --> 277</section> 278 279<section class="section"> 280 <div class="container is-max-widescreen"> 281 <div class="rows"> 282 <div class="rows is-centered"> 283 <div class="row is-full-width"> 284 <h2 class="title is-3 has-text-centered">Rollout Canonical Visualizations</h2> 285 <div id="canonical" class="container" style="display: block;"> 286 <h2 class="subtitle is-5" style="text-align:center;">Given synchronized left and right camera views, we compute a canonical reprojection that is passed to downstream visuomotor policies as a viewpoint-invariant visual representation.</h2> 287 <p><strong>Put Marker In Cup</strong></p> 288 <div class="video_grid_container"> <!-- This is the new container for the flex display --> 289 <div class="video_row"> 290 <div class="video_grid"> 291 <div class="video_grid_item"><div class="vsc-controller"></div> 292 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 293 <source src="videos/Canonical/marker_left.mp4" type="video/mp4"> 294 </video> 295 </div> 296 <p>Left Camera</p> 297 </div> 298 <div class="video_grid"> 299 <div class="video_grid_item"><div class="vsc-controller"></div> 300 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 301 <source src="videos/Canonical/marker_right.mp4" type="video/mp4"> 302 </video> 303 </div> 304 <p>Right Camera</p> 305 </div> 306 <div class="video_grid"> 307 <div class="video_grid_item"><div class="vsc-controller"></div> 308 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 309 <source src="videos/Canonical/marker_canonical.mp4" type="video/mp4"> 310 </video> 311 </div> 312 <p>Canonical Rendering</p> 313 </div> 314 </div> 315 <div class="video-divider"></div> 316 317 </div> 318 <p><strong>Unfold Towel</strong></p> 319 <div class="video_grid_container"> <!-- This is the new container for the flex display --> 320 <div class="video_row"> 321 <div class="video_grid"> 322 <div class="video_grid_item"><div class="vsc-controller"></div> 323 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 324 <source src="videos/Canonical/towel_left.mp4" type="video/mp4"> 325 </video> 326 </div> 327 <p>Left Camera</p> 328 </div> 329 <div class="video_grid"> 330 <div class="video_grid_item"><div class="vsc-controller"></div> 331 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 332 <source src="videos/Canonical/towel_right.mp4" type="video/mp4"> 333 </video> 334 </div> 335 <p>Right Camera</p> 336 </div> 337 <div class="video_grid"> 338 <div class="video_grid_item"><div class="vsc-controller"></div> 339 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 340 <source src="videos/Canonical/towel_canonical.mp4" type="video/mp4"> 341 </video> 342 </div> 343 <p>Canonical Rendering</p> 344 </div> 345 </div> 346 <div class="video-divider"></div> 347 348 </div> 349 <p><strong>Put Brush in Bowl</strong></p> 350 <div class="video_grid_container"> <!-- This is the new container for the flex display --> 351 <div class="video_row"> 352 <div class="video_grid"> 353 <div class="video_grid_item"><div class="vsc-controller"></div> 354 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 355 <source src="videos/Canonical/brush_left.mp4" type="video/mp4"> 356 </video> 357 </div> 358 <p>Left Camera</p> 359 </div> 360 <div class="video_grid"> 361 <div class="video_grid_item"><div class="vsc-controller"></div> 362 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 363 <source src="videos/Canonical/brush_right.mp4" type="video/mp4"> 364 </video> 365 </div> 366 <p>Right Camera</p> 367 </div> 368 <div class="video_grid"> 369 <div class="video_grid_item"><div class="vsc-controller"></div> 370 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 371 <source src="videos/Canonical/brush_canonical.mp4" type="video/mp4"> 372 </video> 373 </div> 374 <p>Canonical Rendering</p> 375 </div> 376 </div> 377 <div class="video-divider"></div> 378 379 </div> 380 <p><strong>Stack Blocks</strong></p> 381 <div class="video_grid_container"> <!-- This is the new container for the flex display --> 382 <div class="video_row"> 383 <div class="video_grid"> 384 <div class="video_grid_item"><div class="vsc-controller"></div> 385 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 386 <source src="videos/Canonical/stack_blocks_left.mp4" type="video/mp4"> 387 </video> 388 </div> 389 <p>Left Camera</p> 390 </div> 391 <div class="video_grid"> 392 <div class="video_grid_item"><div class="vsc-controller"></div> 393 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 394 <source src="videos/Canonical/stack_blocks_right.mp4" type="video/mp4"> 395 </video> 396 </div> 397 <p>Right Camera</p> 398 </div> 399 <div class="video_grid"> 400 <div class="video_grid_item"><div class="vsc-controller"></div> 401 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 402 <source src="videos/Canonical/stack_blocks_canonical.mp4" type="video/mp4"> 403 </video> 404 </div> 405 <p>Canonical Rendering</p> 406 </div> 407 </div> 408 409 </div> 410 </div> 411 </div> 412 </div> 413 </div> 414 </div> <!-- rows --> 415</section> 416 417<section class="section"> 418 <div class="container is-max-widescreen"> 419 <div class="rows"> 420 <div class="rows is-centered"> 421 <div class="row is-full-width"> 422 <h2 class="title is-3 has-text-centered">Qualitative Results</h2> 423 424 <div id="tabBox1"> 425 <div class="tabs is-boxed is-centered"> 426 <ul> 427 <li id="clickDefault" class="tablinks is-active" onclick="openTab(event, 'diverse', 'tabBox1')"><a>Diverse Viewpoint Training Data</a></li> 428 <li class="tablinks" onclick="openTab(event, 'fixed', 'tabBox1')"><a>Limited viewpoint Training Data</a></li> 429 </ul> 430 </div> 431 <div id="diverse" class="container tabcontent" style="display: block;"> 432 <h2 class="subtitle is-5" style="text-align:center;">The following policies were trained on datasets containing substantial variation in camera placement.</h2> 433 <div class="video_grid_container2"> <!-- This is the new container for the flex display --> 434 <p><strong>Put Marker In Cup</strong></p> 435 <div class="video_row2"> 436 <div class="video_grid"> 437 <div class="video_grid_item"><div class="vsc-controller"></div> 438 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 439 <source src="videos/Div/marker_argus_div.mp4" type="video/mp4"> 440 </video> 441 </div> 442 <p>ARGUS + Diffusion Policy</p> 443 </div> 444 <div class="video_grid"> 445 <div class="video_grid_item"><div class="vsc-controller"></div> 446 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 447 <source src="videos/Div/marker_dp_div.mp4" type="video/mp4"> 448 </video> 449 </div> 450 <p>Classic Diffusion Policy</p> 451 </div> 452 <div class="video_grid"> 453 <div class="video_grid_item"><div class="vsc-controller"></div> 454 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 455 <source src="videos/Div/marker_kyc_div.mp4" type="video/mp4"> 456 </video> 457 </div> 458 <p>KYC + Diffusion Policy</p> 459 </div> 460 </div> 461 <div class="video-divider"></div> 462 <p><strong>Unfold Towel</strong></p> 463 <div class="video_row2"> 464 <div class="video_grid"> 465 <div class="video_grid_item"><div class="vsc-controller"></div> 466 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 467 <source src="videos/Div/towel_argus_div.mp4" type="video/mp4"> 468 </video> 469 </div> 470 <p>ARGUS + Diffusion Policy</p> 471 </div> 472 <div class="video_grid"> 473 <div class="video_grid_item"><div class="vsc-controller"></div> 474 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 475 <source src="videos/Div/towel_dp_div.mp4" type="video/mp4"> 476 </video> 477 </div> 478 <p>Classic Diffusion Policy</p> 479 </div> 480 <div class="video_grid"> 481 <div class="video_grid_item"><div class="vsc-controller"></div> 482 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 483 <source src="videos/Div/towel_kyc_div.mp4" type="video/mp4"> 484 </video> 485 </div> 486 <p>KYC + Diffusion Policy</p> 487 </div> 488 </div> 489 <div class="video-divider"></div> 490 <p><strong>Put Brush In Bowl</strong></p> 491 <div class="video_row2"> 492 <div class="video_grid"> 493 <div class="video_grid_item"><div class="vsc-controller"></div> 494 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 495 <source src="videos/Div/brush_argus_div.mp4" type="video/mp4"> 496 </video> 497 </div> 498 <p>ARGUS + Diffusion Policy</p> 499 </div> 500 <div class="video_grid"> 501 <div class="video_grid_item"><div class="vsc-controller"></div> 502 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 503 <source src="videos/Div/brush_dp_div.mp4" type="video/mp4"> 504 </video> 505 </div> 506 <p>Classic Diffusion Policy</p> 507 </div> 508 <div class="video_grid"> 509 <div class="video_grid_item"><div class="vsc-controller"></div> 510 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 511 <source src="videos/Div/brush_kyc_div.mp4" type="video/mp4"> 512 </video> 513 </div> 514 <p>KYC + Diffusion Policy</p> 515 </div> 516 </div> 517 <div class="video-divider"></div> 518 <p><strong>Stack Blocks</strong></p> 519 <div class="video_row2"> 520 <div class="video_grid"> 521 <div class="video_grid_item"><div class="vsc-controller"></div> 522 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 523 <source src="videos/Div/stack_blocks_argus_div.mp4" type="video/mp4"> 524 </video> 525 </div> 526 <p>ARGUS + Diffusion Policy</p> 527 </div> 528 <div class="video_grid"> 529 <div class="video_grid_item"><div class="vsc-controller"></div> 530 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 531 <source src="videos/Div/stack_blocks_dp_div.mp4" type="video/mp4"> 532 </video> 533 </div> 534 <p>Classic Diffusion Policy</p> 535 </div> 536 <div class="video_grid"> 537 <div class="video_grid_item"><div class="vsc-controller"></div> 538 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 539 <source src="videos/Div/stack_blocks_kyc_div.mp4" type="video/mp4"> 540 </video> 541 </div> 542 <p>KYC + Diffusion Policy</p> 543 </div> 544 </div> 545 </div> 546 </div> 547 548 549 <div id="fixed" class="container tabcontent" style="display: none;"> 550 551 <h2 class="subtitle is-5" style="text-align:center;">The following policies were trained on datasets collected from a single fixed multi-view camera configuration.</h2> 552 <div class="video_grid_container2"> <!-- This is the new container for the flex display --> 553 <p><strong>Put Marker In Cup</strong></p> 554 <div class="video_row2"> 555 <div class="video_grid"> 556 <div class="video_grid_item"><div class="vsc-controller"></div> 557 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 558 <source src="videos/ZS/marker_argus_zs.mp4" type="video/mp4"> 559 </video> 560 </div> 561 <p>ARGUS + Diffusion Policy</p> 562 </div> 563 <div class="video_grid"> 564 <div class="video_grid_item"><div class="vsc-controller"></div> 565 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" h
565eight="40%"> 566 <source src="videos/ZS/marker_dp_zs.mp4" type="video/mp4"> 567 </video> 568 </div> 569 <p>Classic Diffusion Policy</p> 570 </div> 571 <div class="video_grid"> 572 <div class="video_grid_item"><div class="vsc-controller"></div> 573 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 574 <source src="videos/ZS/marker_vista_zs.mp4" type="video/mp4"> 575 </video> 576 </div> 577 <p>VISTA + Diffusion Policy</p> 578 </div> 579 </div> 580 <div class="video-divider"></div> 581 <p><strong>Unfold Towel</strong></p> 582 <div class="video_row2"> 583 <div class="video_grid"> 584 <div class="video_grid_item"><div class="vsc-controller"></div> 585 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 586 <source src="videos/ZS/towel_argus_zs.mp4" type="video/mp4"> 587 </video> 588 </div> 589 <p>ARGUS + Diffusion Policy</p> 590 </div> 591 <div class="video_grid"> 592 <div class="video_grid_item"><div class="vsc-controller"></div> 593 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 594 <source src="videos/ZS/towel_dp_zs.mp4" type="video/mp4"> 595 </video> 596 </div> 597 <p>Classic Diffusion Policy</p> 598 </div> 599 <div class="video_grid"> 600 <div class="video_grid_item"><div class="vsc-controller"></div> 601 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 602 <source src="videos/ZS/towel_vista_zs.mp4" type="video/mp4"> 603 </video> 604 </div> 605 <p>VISTA + Diffusion Policy</p> 606 </div> 607 </div> 608 <div class="video-divider"></div> 609 <p><strong>Put Brush In Bowl</strong></p> 610 <div class="video_row2"> 611 <div class="video_grid"> 612 <div class="video_grid_item"><div class="vsc-controller"></div> 613 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 614 <source src="videos/ZS/brush_argus_zs.mp4" type="video/mp4"> 615 </video> 616 </div> 617 <p>ARGUS + Diffusion Policy</p> 618 </div> 619 <div class="video_grid"> 620 <div class="video_grid_item"><div class="vsc-controller"></div> 621 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 622 <source src="videos/ZS/brush_dp_zs.mp4" type="video/mp4"> 623 </video> 624 </div> 625 <p>Classic Diffusion Policy</p> 626 </div> 627 <div class="video_grid"> 628 <div class="video_grid_item"><div class="vsc-controller"></div> 629 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 630 <source src="videos/ZS/brush_vista_zs.mp4" type="video/mp4"> 631 </video> 632 </div> 633 <p>VISTA + Diffusion Policy</p> 634 </div> 635 </div> 636 <div class="video-divider"></div> 637 <p><strong>Stack Blocks</strong></p> 638 <div class="video_row2"> 639 <div class="video_grid"> 640 <div class="video_grid_item"><div class="vsc-controller"></div> 641 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 642 <source src="videos/ZS/stack_blocks_argus_zs.mp4" type="video/mp4"> 643 </video> 644 </div> 645 <p>ARGUS + Diffusion Policy</p> 646 </div> 647 <div class="video_grid"> 648 <div class="video_grid_item"><div class="vsc-controller"></div> 649 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 650 <source src="videos/ZS/stack_blocks_dp_zs.mp4" type="video/mp4"> 651 </video> 652 </div> 653 <p>Classic Diffusion Policy</p> 654 </div> 655 <div class="video_grid"> 656 <div class="video_grid_item"><div class="vsc-controller"></div> 657 <video class="grid-video training" autoplay="" muted="" loop="" playsinline="" height="40%"> 658 <source src="videos/ZS/stack_blocks_vista_zs.mp4" type="video/mp4"> 659 </video> 660 </div> 661 <p>VISTA + Diffusion Policy</p> 662 </div> 663 </div> 664 </div> 665 666 </div> 667 668 669 </div> 670 </div> 671 </div> 672 </div> 673 </div> <!-- rows --> 674</section> 675 676 677 678<section class="section" id="BibTeX"> 679 <div class="container is-max-desktop content"> 680 <h2 class="title">BibTeX</h2> 681 <pre><code>@misc{sathua2026argusaligningrobotscene, 682 title={ARGUS: Aligning Robot Scene Geometry Under Shifting Views with Large 3D Vision Models}, 683 author={Rishik Sathua and Haonan Chen and Katherine Driggs-Campbell}, 684 year={2026}, 685 eprint={2608.05579}, 686 archivePrefix={arXiv}, 687 primaryClass={cs.RO}, 688 url={https://arxiv.org/abs/2608.05579}, 689 }</code></pre> 690 </div> 691</section> 692 693<footer class="footer"> 694 <div class="container"> 695 <div class="columns is-centered"> 696 <div class="column"> 697 <div class="content has-text-centered"> 698 <p> 699 Website template borrowed from <a href="https://github.com/nerfies/nerfies.github.io">NeRFies</a> and <a href="https://haonan16.github.io/stow_page/">Stow</a>. 700 </p> 701 </div> 702 </div> 703 </div> 704 </div> 705</footer> 706 707</body> 708
709<script> 710 function resampleTrajectory() { 711 const demoIdx = Math.floor(Math.random() * 100); // 0 to 99 inclusive 712 const base = `website_data/put_marker_in_cup/put_marker_in_cup/demo_${demoIdx}`; 713 714 const videoSpecs = [ 715 { videoId: 'sample-cam0', srcId: 'sample-cam0-src', file: 'cam_0_rgb.mp4' }, 716 { videoId: 'sample-cam1', srcId: 'sample-cam1-src', file: 'cam_1_rgb.mp4' }, 717 { videoId: 'sample-canonical', srcId: 'sample-canonical-src', file: 'canonical_rgb.mp4' }, 718 ]; 719 720 videoSpecs.forEach(({ videoId, srcId, file }) => { 721 const video = document.getElementById(videoId); 722 const source = document.getElementById(srcId); 723 source.src = `${base}/${file}`; 724 video.load(); 725 video.play(); 726 }); 727 } 728 729 // Optionally load an initial random sample on page load 730 document.addEventListener('DOMContentLoaded', resampleTrajectory); 731</script>
731 732</html>
Line numbers count LF bytes from the start of the resource, as the search results do. Vendor segments are library code the classifier recognised; they are stored but not indexed. Bytes are shown as Latin1 characters, one per byte.