-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathindex.html
More file actions
853 lines (732 loc) · 33.2 KB
/
index.html
File metadata and controls
853 lines (732 loc) · 33.2 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
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
<!DOCTYPE html>
<html>
<head>
<meta charset="utf-8">
<meta name="description"
content="BIMgent: Towards Autonomous Building Modeling via Computer-use Agents">
<meta name="keywords" content="BIM, GUI agents, Multi-agent framework">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title>BIMgent: Towards Autonomous Building Modeling via Computer-use Agents</title>
<link href="https://fonts.googleapis.com/css?family=Google+Sans|Noto+Sans|Castoro"
rel="stylesheet">
<link rel="stylesheet" href="./static/css/bulma.min.css">
<link rel="stylesheet" href="./static/css/bulma-carousel.min.css">
<link rel="stylesheet" href="./static/css/bulma-slider.min.css">
<link rel="stylesheet" href="./static/css/fontawesome.all.min.css">
<link rel="stylesheet"
href="https://cdn.jsdelivr.net/gh/jpswalsh/academicons@1/css/academicons.min.css">
<link rel="stylesheet" href="./static/css/index.css">
<link rel="icon" href="./static/images/bimgenticon.svg">
<script src="https://ajax.googleapis.com/ajax/libs/jquery/3.5.1/jquery.min.js"></script>
<script defer src="./static/js/fontawesome.all.min.js"></script>
<script src="./static/js/bulma-carousel.min.js"></script>
<script src="./static/js/bulma-slider.min.js"></script>
<script src="./static/js/index.js"></script>
</head>
<body>
<nav class="navbar" role="navigation" aria-label="main navigation">
<div class="navbar-brand">
<a role="button" class="navbar-burger" aria-label="menu" aria-expanded="false">
<span aria-hidden="true"></span>
<span aria-hidden="true"></span>
<span aria-hidden="true"></span>
</a>
</div>
<div class="navbar-menu">
<div class="navbar-start" style="flex-grow: 1; justify-content: center;">
<a class="navbar-item" href="https://www.cee.ed.tum.de/ccbe/home/">
<span class="icon">
<i class="fas fa-home"></i>
</span>
</a>
<div class="navbar-item has-dropdown is-hoverable">
<a class="navbar-link">
More Research
</a>
<div class="navbar-dropdown">
<a class="navbar-item" href="https://github.com/dcy0577/Text2BIM">
Text2BIM
</a>
<a class="navbar-item" href="https://github.com/dcy0577/BIM-Command-Recommendation">
Predictive Modeling
</a>
</div>
</div>
</div>
</div>
</nav>
<section class="hero">
<div class="hero-body">
<div class="container is-max-desktop">
<div class="columns is-centered">
<div class="column has-text-centered">
<h1 class="title is-1 publication-title">BIMgent: Towards Autonomous Building Modeling via Computer-use Agents</h1>
<div class="has-text-centered" style="margin-bottom: 1.5rem;">
<span style="font-size: 1.2rem; font-family: 'Inter', 'Segoe UI', Arial, sans-serif; font-weight: 500; color: black;">
ICML 2025 Workshop on Computer Use Agents
</span>
</div>
<div class="is-size-5 publication-authors">
<span class="author-block">
<a href="https://scholar.google.com/citations?user=u4iSbGwAAAAJ&hl=en">Zihan Deng</a><sup>1,2</sup>,</span>
<span class="author-block">
<a href="https://scholar.google.com/citations?user=ZdNOee8AAAAJ&hl=en">Changyu Du</a><sup>1,2</sup>,</span>
<span class="author-block">
<a href="https://scholar.google.com/citations?user=HQuAxd0AAAAJ&hl=en">Stavros Nousias</a><sup>1,2</sup>,
</span>
<span class="author-block">
<a href="https://scholar.google.com/citations?user=9LSpz44AAAAJ&hl=en">André Borrmann</a><sup>1,2</sup>,
</span>
</div>
<div class="is-size-5 publication-authors">
<span class="author-block"><sup>1</sup>Chair of Computing in Civil and Building Engineering, Technical University of Munich</span>
<span class="author-block"><sup>2</sup>TUM Georg Nemetschek Institute</span>
</div>
<div class="column has-text-centered">
<div class="publication-links">
<span class="link-block">
<a href="https://arxiv.org/abs/2506.07217"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="ai ai-arxiv"></i>
</span>
<span>arXiv</span>
</a>
</span>
<!-- Code Link. -->
<span class="link-block">
<!-- <a class="external-link button is-normal is-rounded is-dark is-disabled" style="background-color: #808080; cursor: not-allowed;"> -->
<a class="external-link button is-normal is-rounded is-dark"
href="https://github.com/ZihanDDD/BIMgent">
<span class="icon">
<i class="fab fa-github"></i>
</span>
<span>Code</span>
</a>
</span>
</div>
</div>
</div>
</div>
</div>
</div>
</section>
<section class="hero teaser" style="margin-top: -1rem;">
<div class="container is-max-desktop">
<div class="hero-body">
<img src="./static/BIMgent_images/general_workflow1.png" height="100%">
<h2 class="subtitle has-text-centered">
We present <span class="dnerf">BIMgent</span>, the first LLM-powered agentic framework to explore autonomous building modeling through computer control.
</h2>
</div>
</div>
</section>
<section class="hero is-light is-small">
<div class="hero-body">
<div class="container is-fluid wide-container">
<h2 class="subtitle has-text-centered" style="margin-bottom: 3rem;">
<b>BIMgent</b> generates BIM model in Vectorworks via GUI operations.
</h2>
<div class="video-list">
<!-- column titles -->
<div class="columns is-vcentered video-header">
<div class="column task-input"><h4 class="title is-6 has-text-centered">User's Input</h4></div>
<div class="column media-col"><h4 class="title is-6 has-text-centered">Modeling Process</h4></div>
<div class="column media-col result-col"><h4 class="title is-6 has-text-centered">Result</h4></div>
</div>
<!-- task1 -->
<div class="columns is-vcentered video-gif-triplet">
<div class="column task-input">
<p>Generate a one-storey office building with a large open workspace occupying most of the floor area. The layout should also include two enclosed meeting rooms, a manager’s office, a small pantry, and two restrooms.</p>
</div>
<div class="column media-col">
<video autoplay controls muted loop playsinline style="width:100%;height:auto;">
<source src="./static/BIMgent_videos/Full_videos/task1.mp4" type="video/mp4">
</video>
</div>
<div class="column media-col result-col">
<img src="./static/BIMgent_gifs/task1.gif" alt="task1 GIF" style="width:100%;height:auto;object-fit:contain;">
</div>
</div>
<!-- task3 -->
<div class="columns is-vcentered video-gif-triplet">
<div class="column task-input">
<p>Create a two-storey rectangular residential building designed for a single family. The layout must include a living room, kitchen, bathroom, master bedroom, and one additional bedroom.</p>
</div>
<div class="column media-col">
<video autoplay controls muted loop playsinline style="width:100%;height:auto;">
<source src="./static/BIMgent_videos/Full_videos/task3.mp4" type="video/mp4">
</video>
</div>
<div class="column media-col result-col">
<img src="./static/BIMgent_gifs/task3.gif" alt="task3 GIF" style="width:100%;height:auto;object-fit:contain;">
</div>
</div>
<!-- task6 -->
<div class="columns is-vcentered video-gif-triplet">
<div class="column task-input">
<p>Generate a one-storey octagonal building based on the hand-drawn sketch:</p>
<img src="./static/BIMgent_images/prompt_06.png" alt="Prompt for task6" class="prompt-img">
</div>
<div class="column media-col">
<video autoplay controls muted loop playsinline style="width:100%;height:auto;">
<source src="./static/BIMgent_videos/Full_videos/task6.mp4" type="video/mp4">
</video>
</div>
<div class="column media-col result-col">
<img src="./static/BIMgent_gifs/task6.gif" alt="task6 GIF" style="width:100%;height:auto;object-fit:contain;">
</div>
</div>
<!-- task12 -->
<div class="columns is-vcentered video-gif-triplet">
<div class="column task-input">
<p>Generate a one-floor building based on the image:</p>
<img src="./static/BIMgent_images/prompt_012.png" alt="Prompt for task12" class="prompt-img">
</div>
<div class="column media-col">
<video autoplay controls muted loop playsinline style="width:100%;height:auto;">
<source src="./static/BIMgent_videos/Full_videos/task12.mp4" type="video/mp4">
</video>
</div>
<div class="column media-col result-col">
<img src="./static/BIMgent_gifs/task12.gif" alt="task12 GIF" style="width:100%;height:auto;object-fit:contain;">
</div>
</div>
<!-- task16 -->
<div class="columns is-vcentered video-gif-triplet">
<div class="column task-input">
<p>Generate a building model based on a hand-drawn octagon floorplan, modifying the interior layout to include four rooms instead of three:</p>
<img src="./static/BIMgent_images/prompt_06.png" alt="Prompt for task16" class="prompt-img">
</div>
<div class="column media-col">
<video autoplay controls muted loop playsinline style="width:100%;height:auto;">
<source src="./static/BIMgent_videos/Full_videos/task16.mp4" type="video/mp4">
</video>
</div>
<div class="column media-col result-col">
<img src="./static/BIMgent_gifs/task16.gif" alt="task16 GIF" style="width:100%;height:auto;object-fit:contain;">
</div>
</div>
</div>
</div>
</div>
</section>
<style>
@media screen and (max-width: 1023px) {
.navbar-brand {
width: 100%;
display: flex;
justify-content: center;
}
.navbar-burger {
margin: 0; /* 移除默认的 margin */
position: relative;
}
/* 确保展开的菜单也居中 */
.navbar-menu.is-active {
text-align: center;
}
.navbar-menu.is-active .navbar-start {
justify-content: center;
}
.navbar-menu.is-active .navbar-item {
display: inline-block;
}
}
/* widen overall container */
.wide-container { max-width: 1440px; margin-inline: auto; }
/* header row */
.video-header { margin-bottom: .75rem; }
/* data rows */
.video-gif-triplet {
margin-bottom: 1.5rem;
padding-bottom: 1.5rem;
border-bottom: 1px solid #e5e5e5;
}
/* column widths: 30% · 40% · 30% */
.task-input { flex: 0 0 30%; }
.media-col:not(.result-col) { flex: 0 0 40%; }
.media-col.result-col { flex: 0 0 30%; }
/* text styles */
.task-input p { font-size: .95rem; line-height: 1.4; font-weight: 700; }
.video-header .task-input h4{ font-weight: 700; }
/* prompt image */
.task-input .prompt-img {
display: block;
width: 100%;
height: 120px;
object-fit: contain;
margin-top: .5rem;
}
/* 去除竖向滚动条 */
.hero.is-light .hero-body {
overflow-y: visible; /* 或者 auto，不要用 hidden */
overflow-x: hidden; /* 只隐藏横向溢出 */
}
/* 确保视频高度自适应 */
.video-gif-triplet video {
max-height: 400px; /* 限制最大高度，避免过高 */
width: 100%;
height: auto;
object-fit: contain;
}
/* 确保 GIF 也有合理高度 */
.video-gif-triplet img {
max-height: 400px;
width: 100%;
height: auto;
object-fit: contain;
}
/* 移动端优化 */
@media (max-width: 768px) {
.video-gif-triplet video,
.video-gif-triplet img {
max-height: 250px; /* 移动端更小的高度 */
}
}
.hero.is-light.is-small .hero-body {
min-height: auto; /* 移除最小高度限制 */
padding-top: 3rem;
padding-bottom: 3rem;
}
/* 针对这个特定容器移除滚动条 */
.container.is-fluid.wide-container {
overflow: visible !important;
max-height: none !important;
height: auto !important;
}
@media screen and (max-width: 768px) {
/* 隐藏表头 */
.video-header {
display: none;
}
/* 改变每个任务的布局 */
.video-gif-triplet {
flex-direction: column;
}
.video-gif-triplet .column {
flex: none !important;
width: 100% !important;
max-width: 100% !important;
margin-bottom: 1rem;
}
/* 在每个内容前添加标签 */
.video-gif-triplet .task-input::before {
content: "User's Input";
display: block;
font-weight: bold;
margin-bottom: 0.5rem;
color: #666;
text-align: center;
}
.video-gif-triplet .media-col:not(.result-col)::before {
content: "Modeling Process";
display: block;
font-weight: bold;
margin-bottom: 0.5rem;
color: #666;
text-align: center;
}
.video-gif-triplet .result-col::before {
content: "Result";
display: block;
font-weight: bold;
margin-bottom: 0.5rem;
color: #666;
text-align: center;
}
}
</style>
<section class="section">
<div class="container is-max-desktop">
<!-- Abstract. -->
<div class="columns is-centered has-text-centered">
<div class="column is-four-fifths">
<h2 class="title is-3">Abstract</h2>
<div class="content has-text-justified">
<p>
Existing computer-use agents primarily focus on general-purpose desktop automation tasks, with limited exploration of their application in highly specialized domains. In particular, the 3D building modeling process in the Architecture, Engineering, and Construction (AEC) sector involves open-ended design tasks and complex interaction patterns within Building Information Modeling (BIM) authoring software, which has yet to be thoroughly addressed by current studies.
</p>
<p>
In this study, we propose <b>BIMgent</b>, an agentic framework powered by multimodal large language models (LLMs), designed to enable autonomous building model authoring via graphical user interface (GUI) operations. BIMgent automates the architectural building modeling process, including multimodal input for conceptual design, planning of software-specific workflows, and efficient execution of the authoring GUI actions.
</p>
<p>
We evaluate BIMgent on real-world building modeling tasks, including both text-based conceptual design generation and reconstruction from existing building design. The design quality achieved by BIMgent was found to be reasonable. Its operations achieved a 32% success rate, whereas all baseline models failed to complete the tasks (0% success rate). Results demonstrate that BIMgent effectively reduces manual workload while preserving design intent, highlighting its potential for practical deployment in real-world architectural modeling scenarios.
</p>
</div>
</div>
</div>
<!--/ Abstract. -->
</div>
<!--/ Paper video. -->
</div>
</section>
<!-- BIMgent decomposed-actions slider -->
<link rel="stylesheet" href="https://unpkg.com/swiper/swiper-bundle.min.css" />
<section class="hero is-light is-small">
<div class="hero-body">
<div class="container">
<h2 class="title is-3 has-text-centered">Execution trajectories</h2>
<div class="swiper-container decomposed-slider">
<div class="swiper-wrapper">
<!-- slide 1 -->
<div class="swiper-slide video-card">
<h3 class="subtitle is-5 has-text-centered">User Input</h3>
<video autoplay controls muted loop playsinline>
<source src="./static/BIMgent_videos/decomposed_videos/users_input.mp4" type="video/mp4">
</video>
</div>
<!-- slide 2 -->
<div class="swiper-slide video-card">
<h3 class="subtitle is-5 has-text-centered">Floorplan Design</h3>
<video autoplay controls muted loop playsinline>
<source src="./static/BIMgent_videos/decomposed_videos/design.mp4" type="video/mp4">
</video>
</div>
<!-- slide 3 -->
<div class="swiper-slide video-card">
<h3 class="subtitle is-5 has-text-centered">Layer Creation</h3>
<video autoplay controls muted loop playsinline>
<source src="./static/BIMgent_videos/decomposed_videos/layer_creation.mp4" type="video/mp4">
</video>
</div>
<!-- slide 4 -->
<div class="swiper-slide video-card">
<h3 class="subtitle is-5 has-text-centered">External Walls Creation</h3>
<video autoplay controls muted loop playsinline>
<source src="./static/BIMgent_videos/decomposed_videos/External_wall.mp4" type="video/mp4">
</video>
</div>
<!-- slide 5 -->
<div class="swiper-slide video-card">
<h3 class="subtitle is-5 has-text-centered">Slab Creation</h3>
<video autoplay controls muted loop playsinline>
<source src="./static/BIMgent_videos/decomposed_videos/slab_creation.mp4" type="video/mp4">
</video>
</div>
<!-- slide 6 -->
<div class="swiper-slide video-card">
<h3 class="subtitle is-5 has-text-centered">Internal Walls Creation</h3>
<video autoplay controls muted loop playsinline>
<source src="./static/BIMgent_videos/decomposed_videos/Internal_wall.mp4" type="video/mp4">
</video>
</div>
<!-- slide 7 -->
<div class="swiper-slide video-card">
<h3 class="subtitle is-5 has-text-centered">Windows Creation</h3>
<video autoplay controls muted loop playsinline>
<source src="./static/BIMgent_videos/decomposed_videos/window.mp4" type="video/mp4">
</video>
</div>
<!-- slide 8 -->
<div class="swiper-slide video-card">
<h3 class="subtitle is-5 has-text-centered">Doors Creation</h3>
<video autoplay controls muted loop playsinline>
<source src="./static/BIMgent_videos/decomposed_videos/door.mp4" type="video/mp4">
</video>
</div>
<!-- slide 9 -->
<div class="swiper-slide video-card">
<h3 class="subtitle is-5 has-text-centered">Roof Creation</h3>
<video autoplay controls muted loop playsinline>
<source src="./static/BIMgent_videos/decomposed_videos/Roof.mp4" type="video/mp4">
</video>
</div>
</div>
<!-- navigation -->
<div class="swiper-button-prev"></div>
<div class="swiper-button-next"></div>
<div class="swiper-pagination"></div>
</div>
</div>
</div>
</section>
<style>
/* 更大视频尺寸 */
.decomposed-slider .video-card{
width:100%;
box-sizing:border-box;
padding:0 1rem;
}
.decomposed-slider video{
width:100%;
height:480px; /* 高度再提升 */
border-radius:10px;
object-fit:contain;
}
.decomposed-slider .subtitle{
margin-bottom:0.8rem; /* 标题改在上面，用 margin-bottom */
font-weight:600;
}
.swiper-container{
width:100%;
padding-bottom:2.5rem;
}
/* 让标题排在视频上方 */
.decomposed-slider .video-card{
width:100%;
box-sizing:border-box;
padding:0 1rem;
/* 新增 ↓↓↓ */
display:flex; /* 让子元素走 Flex 布局 */
flex-direction:column; /* 主轴改成竖直方向，标题自然在上 */
align-items:center; /* 可选：让视频水平居中 */
}
.decomposed-slider .subtitle{
width:100%; /* 占满整行，避免被挤到侧面 */
text-align:center;
margin-bottom:0.8rem;
font-weight:600;
}
/* 修复横向溢出问题 - 在父容器上设置 */
.hero.is-light .hero-body {
overflow-y: visible;
overflow-x: hidden; /* 防止横向滚动 */
}
.hero.is-light .container {
overflow-x: hidden; /* 确保容器不会横向溢出 */
}
/* 限制 Swiper 容器宽度 */
.swiper-container.decomposed-slider {
max-width: 100%;
margin: 0 auto;
}
/* 移动端优化 */
@media (max-width: 768px) {
.hero.is-light {
overflow-x: hidden;
}
.decomposed-slider .video-card {
padding: 0 0.5rem;
}
/* 确保视频不会溢出 */
.decomposed-slider video {
max-width: 100%;
height: auto;
max-height: 480px;
}
}
</style>
<script src="https://unpkg.com/swiper/swiper-bundle.min.js"></script>
<script>
new Swiper('.decomposed-slider',{
slidesPerView:1, // 每屏只显示 1 个卡片
spaceBetween:32,
watchOverflow: true,
navigation:{
nextEl:'.swiper-button-next',
prevEl:'.swiper-button-prev'
},
pagination:{
el:'.swiper-pagination',
clickable:true
},
breakpoints:{
0:{slidesPerView:1, spaceBetween:20}, // 移动端减少间距
768:{slidesPerView:1, spaceBetween:32},
1024:{slidesPerView:1, spaceBetween:32}
}
});
</script>
<script src="https://unpkg.com/swiper/swiper-bundle.min.js"></script>
<script>
new Swiper('.bimgent-slider',{
slidesPerView:1, // one slide (column) at a time
spaceBetween:40,
navigation:{
nextEl:'.swiper-button-next',
prevEl:'.swiper-button-prev'
},
pagination:{
el:'.swiper-pagination',
clickable:true
}
});
</script>
<!-- ==================== Method Overview ==================== -->
<section id="method" class="section">
<div class="container is-max-desktop">
<div class="columns is-centered">
<div class="column is-full">
<header class="has-text-centered mb-6">
<h2 class="title is-3">Method</h2>
</header>
<div class="content">
<figure class="image mb-5" style="max-width:2000px;margin-inline:auto;">
<img
src="./static/BIMgent_images/detail_workflow.png"
alt="Workflow diagram of the BIMgent framework, showing the Design, Action-Planning, and Execution layers."
style="width:100%;height:auto;display:block;"
loading="lazy"
/>
</figure>
<p>
Overview of the <b>BIMgent</b> framework. Given the multimodal design requirements provided by the user, the <b>Design Layer</b> first transforms them into a refined floorplan and extracts the necessary semantic and geometric information to guide the modeling process. Based on the interpreted design information and domain knowledge, the <b>Action Planning</b> Layer hierarchically organizes the modeling procedure and decomposes it into detailed substeps, guided by the official software documentation. These substeps are then executed through specialized action workflows in the <b>Execution Layer</b>, each equipped with verification mechanisms. Execution trajectories are stored in a memory module, which supports both self-reflection and cooperation among different parts of the framework.
</p>
</div>
</div>
</div>
</div>
</section>
<!-- Experiments section with figure captions (no corner labels) -->
<section id="method" class="section">
<div class="container is-max-desktop">
<div class="columns is-centered">
<div class="column is-full">
<!-- ── Header ──────────────────────────────────────────────────────────── -->
<header class="has-text-centered mb-6">
<h2 class="title is-3">Experiments</h2>
</header>
<!-- ── Content ──────────────────────────────────────────────────────────── -->
<div class="content">
<!-- Mini Building Benchmark -->
<h3 class="title is-4 mb-4">Mini Building Benchmark</h3>
<p>
We present a <strong>Mini Building Benchmark</strong> consisting of <strong>25 BIM authoring
tasks</strong>. These tasks cover five input scenarios:
</p>
<ol>
<li><strong>Conceptual text descriptions</strong> of design intent, detailing the building type, room program, and one- to three-storey height.</li>
<li><strong>Hand-drawn sketch floor plans</strong> illustrating both regular and irregular shapes, with the specified number of floors.</li>
<li><strong>Unmodified floor-plan images</strong> randomly selected from the <a href="https://github.com/CubiCasa/CubiCasa5k" target="_blank">CubiCasa5K</a> dataset.</li>
<li>The same hand-drawn floor plans with <strong>explicit modification requests</strong> (e.g. “add a room”).</li>
<li>The same <a href="https://github.com/CubiCasa/CubiCasa5k" target="_blank">CubiCasa5K</a> floor plans with <strong>explicit modification instructions</strong>.</li>
</ol>
<!-- ── Figures: dataset distribution & design-evaluation criteria ───────── -->
<div class="columns is-centered mb-5">
<div class="column is-half">
<figure class="image is-3by2">
<img
src="./static/BIMgent_images/dataset_distribution.png"
alt="Dataset distribution across the five input scenarios for the Mini Building Benchmark."
loading="lazy"
style="object-fit: contain;"
/>
<figcaption class="has-text-centered is-size-7 mt-2"><b>Task Distribution in Mini Building Benchmark</b></figcaption>
</figure>
</div>
<div class="column is-half">
<figure class="image is-3by2">
<img
src="./static/BIMgent_images/design_eva.png"
alt="Illustration of the six design-evaluation criteria"
style="object-fit: contain;"
/>
<figcaption class="has-text-centered is-size-7 mt-2"><b>Design Evaluation</b></figcaption>
</figure>
</div>
</div>
<!-- Design Evaluation -->
<h3 class="title is-4 mb-3" role="heading" aria-level="3">Design Evaluation</h3>
<p>
We conducted a manual review to evaluate the generated floorplans against six design criteria, comparing our full design layer with two baselines:
</p>
<ol>
<li>SVG floorplans generated by Claude 3.7</li>
<li>Our design layer <em>without</em> the floorplan-interpretation module</li>
</ol>
<p>
The results indicate our method is a promising direction. It consistently scored above <strong>3 / 5</strong> on all criteria and outperformed both baselines. However, we acknowledge that the overall design quality is not yet sufficient for professional use and there remains a significant gap to bridge before the designs meet a standard satisfactory to architects.
</p>
<!-- Operation Evaluation -->
<h3 class="title is-4 mb-3" role="heading" aria-level="3">Operation Evaluation</h3>
<figure class="image mb-5" style="margin: 0 auto;">
<img
src="./static/BIMgent_images/eva.png"
alt="Comparison of design-criterion scores across methods"
style="display: block; width: 100%; height: auto; object-fit: contain;"
/>
</figure>
<!-- Benchmark results block -->
<div class="content">
<p>
On the Mini Building Benchmark, BIMgent achieves a <strong>32 %</strong> end-to-end success rate. Because overall success depends on many intermediate steps, we augmented the 25 complete modeling tasks with subtasks targeting key architectural components:
</p>
<ul>
<li><strong>Design layers</strong> — 41 tasks</li>
<li><strong>Walls</strong> — 82 tasks</li>
<li><strong>Slabs</strong> — 41 tasks</li>
<li><strong>Openings</strong> — 82 tasks</li>
<li><strong>Roofs</strong> — 25 tasks</li>
</ul>
<p>
A subtask is successful only if all required elements are created and their parameters are correctly configured. BIMgent excels at component-level subtasks, achieving <strong>86.58 %</strong> success on walls and <strong>92.68 %</strong> on openings. The strongest baseline (Claude 3.7) failed every end-to-end task due to heavy planning and extensive GUI operations, and performed poorly on all subtasks.
</p>
<p>
Our framework also attains <strong>46.34 %</strong> on layer and <strong>60 %</strong> on roof tasks (vs. <strong>0 %</strong> for the baseline), and significantly outperforms the baseline on walls, slabs, and openings. These results highlight BIMgent's potential to automate tedious modeling operations in professional BIM authoring software.
</p>
</div>
</div> <!-- /content -->
</div> <!-- /column -->
</div> <!-- /columns -->
</div> <!-- /container -->
</section>
<!-- ==================== Conclusion ==================== -->
<section id="conclusion" class="section">
<div class="container is-max-desktop">
<div class="columns is-centered">
<div class="column is-full">
<header class="has-text-centered mb-6">
<h2 class="title is-3">Conclusion</h2>
</header>
<div class="content">
<p>
In this study, we explore an important yet under-explored question: Is it possible to achieve automated modeling in professional design software using GUI agent technology? Although current performance limitations prevent immediate deployment of the proposed approach, this framework lays a foundation for developing domain-specific computer-use agents in high-stakes fields like AEC. The contributions of this study can be summarized as follows:
</p>
<ul>
<li>Compared to existing GUI agents, <strong>BIMgent</strong> demonstrates the ability to handle open-ended design tasks, bridging the gap in applying GUI agents to professional building design software.</li>
<li><strong>BIMgent</strong> addresses the limitations of domain-specific task handling by integrating software documentation into the planning process through a retrieval-augmented strategy.</li>
<li><strong>BIMgent</strong> significantly boosts performance by combining dynamic GUI grounding, reflective feedback, and hierarchical planning, thereby overcoming the challenges of complex GUIs and the hundreds of steps required for building modeling tasks.</li>
<li>By evaluating two stages of the framework, we show that <strong>BIMgent</strong> can complete the entire modeling process autonomously, particularly excelling in the most labor-intensive parts of the modeling workflow.</li>
</ul>
<p>
Future work will focus on extending the framework to other design software to test its generalizability, and on optimizing the agent's step count and execution time to further reduce manual effort. Fine-tuning open-source models may also improve adaptability and performance. Additionally, we plan to develop automated evaluation methods and introduce a dedicated benchmark for more consistent and scalable assessment.
</p>
</div>
</div>
</div>
</div>
</section>
<section class="section" id="BibTeX">
<div class="container is-max-desktop content">
<h2 class="title">BibTeX</h2>
<pre><code>@article{deng2025bimgent,
title={BIMgent: Towards Autonomous Building Modeling via Computer-use Agents},
author={Deng, Zihan and Du, Changyu and Nousias, Stavros and Borrmann, Andr{\'e}},
journal={ICML 2025 Workshop on Computer Use Agents},
year={2025}
}</code></pre>
</div>
</section>
<footer class="footer">
<div class="container">
<!-- <div class="content has-text-centered">
<a class="icon-link"
href="./static/videos/nerfies_paper.pdf">
<i class="fas fa-file-pdf"></i>
</a>
<a class="icon-link" href="" class="external-link" disabled>
<i class="fab fa-github"></i>
</a>
</div> -->
<div class="columns is-centered">
<div class="column is-8">
<div class="content has-text-centered">
<p>
This website template is adapted from source at <a href="https://github.com/nerfies/nerfies.github.io">Nerfies</a>.
</p>
</div>
</div>
</div>
</div>
<!-- Visitor Map Section -->
<div class="columns is-centered">
<div class="column is-8">
<div class="content has-text-centered">
<a href='https://mapmyvisitors.com/web/1byhp' title='Visit tracker'><img src='https://mapmyvisitors.com/map.png?cl=ffffff&w=300&t=n&d=W-f9vQnuf7hEx97PGBb-FuX73qRLU23SuEQbUfqRzL4'/></a>
</div>
</div>
</div>
</footer>
</body>
</html>