| <!DOCTYPE html> |
| <html lang="en"> |
| <head> |
| <meta charset="UTF-8"> |
| <meta name="viewport" content="width=device-width, initial-scale=1.0"> |
| <title>SABER — A Scalable Action-Based Embodied Dataset for Real-World VLA Adaptation | DreamVu</title> |
| <meta name="description" content="SABER: 44.8K robot-training samples from 100+ hours of real in-store human activity. Three complementary action supervision streams for domain-specific VLA adaptation."> |
| <link rel="icon" type="image/x-icon" href="https://dreamvu.ai/wp-content/themes/DreamVU%20Custom%20theme/assets/images/favicon.ico"> |
| <link href="https://fonts.googleapis.com/css2?family=Inter:wght@300;400;500;600;700;800;900&display=swap" rel="stylesheet"> |
| <style> |
| |
| |
| |
| :root { |
| --bg-dark: #0f1419; |
| --bg-darker: #0b1729; |
| --bg-card: #1c2d44; |
| --bg-card-hover: #243550; |
| --orange: #f5a623; |
| --orange-bright: #ffb84d; |
| --blue-accent: #3b82f6; |
| --blue-glow: rgba(59, 130, 246, 0.15); |
| --cyan: #06b6d4; |
| --green: #7BF1A8; |
| --text-primary: #f0f2f5; |
| --text-secondary: #94a3b8; |
| --text-muted: #64748b; |
| --gradient-orange: linear-gradient(135deg, #f5a623, #ff6b35); |
| --gradient-blue: linear-gradient(135deg, #3b82f6, #06b6d4); |
| --border-subtle: rgba(255, 255, 255, 0.06); |
| --border-light: rgba(255, 255, 255, 0.1); |
| } |
| |
| *, *::before, *::after { box-sizing: border-box; margin: 0; padding: 0; } |
| html { scroll-behavior: smooth; overflow-x: hidden; -webkit-text-size-adjust: 100%; } |
| body { |
| font-family: 'Inter', -apple-system, BlinkMacSystemFont, sans-serif; |
| background: var(--bg-dark); |
| color: var(--text-primary); |
| line-height: 1.6; |
| -webkit-font-smoothing: antialiased; |
| overflow-x: hidden; |
| } |
| |
| |
| nav { |
| position: fixed; top: 0; left: 0; width: 100%; z-index: 1000; |
| height: 72px; padding: 0 48px; |
| display: flex; align-items: center; justify-content: space-between; |
| background: rgba(11, 23, 41, 0.85); |
| backdrop-filter: blur(20px); -webkit-backdrop-filter: blur(20px); |
| border-bottom: 1px solid var(--border-subtle); |
| } |
| .nav-logo img { height: 28px; } |
| .nav-links { display: flex; align-items: center; gap: 32px; } |
| .nav-links a { |
| font-size: 14px; font-weight: 500; color: var(--text-secondary); |
| text-decoration: none; transition: color 0.2s; |
| } |
| .nav-links a:hover { color: var(--orange); } |
| .btn-cta { |
| display: inline-flex; align-items: center; gap: 8px; |
| background: var(--gradient-orange); color: #000; |
| padding: 10px 24px; border-radius: 8px; |
| font-size: 14px; font-weight: 600; text-decoration: none; |
| transition: all 0.2s; |
| } |
| .btn-cta:hover { transform: translateY(-1px); box-shadow: 0 8px 24px rgba(245, 166, 35, 0.3); } |
| .hamburger { display: none; background: none; border: none; cursor: pointer; flex-direction: column; gap: 5px; padding: 8px; } |
| .hamburger span { width: 24px; height: 2px; background: var(--text-primary); border-radius: 2px; transition: all 0.3s; } |
| .hamburger.active span:nth-child(1) { transform: rotate(45deg) translate(5px, 5px); } |
| .hamburger.active span:nth-child(2) { opacity: 0; } |
| .hamburger.active span:nth-child(3) { transform: rotate(-45deg) translate(5px, -5px); } |
| .mobile-menu { |
| display: none; position: fixed; top: 72px; left: 0; right: 0; bottom: 0; |
| background: rgba(15, 20, 25, 0.98); backdrop-filter: blur(20px); |
| z-index: 999; padding: 32px 40px; flex-direction: column; gap: 8px; |
| overflow-y: auto; |
| } |
| .mobile-menu.open { display: flex; } |
| .mobile-menu a { |
| font-size: 16px; font-weight: 500; color: var(--text-secondary); |
| text-decoration: none; padding: 16px 0; |
| border-bottom: 1px solid var(--border-subtle); |
| transition: color 0.2s; |
| } |
| .mobile-menu a:hover { color: var(--orange); } |
| |
| |
| .section-inner { max-width: 1200px; margin: 0 auto; } |
| section { padding: 120px 48px 80px; } |
| .section-label { |
| font-size: 12px; font-weight: 600; color: var(--orange); |
| text-transform: uppercase; letter-spacing: 0.15em; margin-bottom: 12px; |
| } |
| .section-title { |
| font-size: 40px; font-weight: 800; letter-spacing: -0.02em; |
| line-height: 1.15; margin-bottom: 16px; |
| } |
| .section-subtitle { |
| font-size: 17px; color: var(--text-secondary); max-width: 720px; |
| line-height: 1.7; margin-bottom: 48px; |
| } |
| |
| |
| .hero { |
| min-height: 100vh; padding: 140px 48px 80px; |
| display: flex; align-items: center; position: relative; overflow: hidden; |
| } |
| .hero::before { |
| content: ''; position: absolute; |
| top: -20%; left: 50%; width: 900px; height: 900px; |
| background: radial-gradient(circle, rgba(59,130,246,0.08) 0%, transparent 70%); |
| transform: translate(-50%, 0); |
| animation: orbFloat 10s infinite alternate ease-in-out; |
| pointer-events: none; |
| } |
| .hero::after { |
| content: ''; position: absolute; |
| bottom: -10%; right: 10%; width: 600px; height: 600px; |
| background: radial-gradient(circle, rgba(245,166,35,0.05) 0%, transparent 70%); |
| pointer-events: none; |
| } |
| @keyframes orbFloat { |
| 0% { transform: translate(-50%, 0) scale(1); } |
| 100% { transform: translate(-45%, -5%) scale(1.1); } |
| } |
| .hero-content { |
| max-width: 1200px; margin: 0 auto; width: 100%; |
| display: grid; grid-template-columns: 1fr 380px; gap: 64px; |
| align-items: start; position: relative; z-index: 2; |
| } |
| .hero-badge { |
| display: inline-flex; align-items: center; gap: 8px; |
| padding: 6px 16px; border-radius: 100px; |
| background: rgba(59, 130, 246, 0.1); border: 1px solid rgba(59, 130, 246, 0.3); |
| font-size: 12px; font-weight: 600; color: var(--blue-accent); |
| margin-bottom: 24px; text-transform: uppercase; letter-spacing: 0.08em; |
| } |
| .hero-badge .dot { |
| width: 8px; height: 8px; border-radius: 50%; |
| background: var(--green); animation: pulse 1.5s infinite; |
| } |
| @keyframes pulse { |
| 0% { transform: scale(0.9); opacity: 0.7; } |
| 50% { transform: scale(1.2); opacity: 1; } |
| 100% { transform: scale(0.9); opacity: 0.7; } |
| } |
| .hero h1 { |
| font-size: 56px; font-weight: 800; line-height: 1.08; |
| letter-spacing: -0.03em; margin-bottom: 8px; |
| } |
| .hero h1 .highlight { |
| background: var(--gradient-orange); |
| -webkit-background-clip: text; -webkit-text-fill-color: transparent; |
| background-clip: text; |
| } |
| .hero .paper-full-title { |
| font-size: 18px; font-weight: 400; color: var(--text-secondary); |
| line-height: 1.6; margin-bottom: 24px; max-width: 560px; |
| } |
| .hero-buttons { display: flex; gap: 12px; flex-wrap: wrap; margin-bottom: 40px; } |
| .btn-outline { |
| display: inline-flex; align-items: center; gap: 8px; |
| padding: 10px 24px; border-radius: 8px; |
| border: 1px solid rgba(255,255,255,0.15); background: transparent; |
| color: var(--text-primary); font-size: 14px; font-weight: 600; |
| text-decoration: none; transition: all 0.2s; |
| } |
| .btn-outline:hover { border-color: var(--orange); color: var(--orange); } |
| .hero-stats { |
| display: grid; grid-template-columns: repeat(3, 1fr); gap: 0; |
| border: 1px solid var(--border-light); border-radius: 16px; overflow: hidden; |
| } |
| .hero-stat { |
| padding: 20px 24px; text-align: center; |
| border-right: 1px solid var(--border-light); |
| background: rgba(255,255,255,0.02); |
| } |
| .hero-stat:last-child { border-right: none; } |
| .hero-stat .number { |
| font-size: 36px; font-weight: 800; |
| background: var(--gradient-orange); |
| -webkit-background-clip: text; -webkit-text-fill-color: transparent; |
| } |
| .hero-stat .label { |
| font-size: 12px; color: var(--text-muted); |
| text-transform: uppercase; letter-spacing: 0.08em; margin-top: 4px; |
| } |
| |
| |
| .resource-stack { display: flex; flex-direction: column; gap: 12px; } |
| .resource-card { |
| display: flex; align-items: center; gap: 16px; |
| padding: 18px 20px; border-radius: 14px; |
| background: var(--bg-card); border: 1px solid var(--border-light); |
| text-decoration: none; color: var(--text-primary); |
| transition: all 0.3s; cursor: pointer; |
| } |
| .resource-card:hover { border-color: rgba(245,166,35,0.2); background: var(--bg-card-hover); transform: translateX(4px); } |
| .resource-icon { |
| width: 42px; height: 42px; border-radius: 10px; |
| display: flex; align-items: center; justify-content: center; flex-shrink: 0; |
| } |
| .icon-orange { background: rgba(245,166,35,0.12); color: var(--orange); } |
| .icon-blue { background: rgba(59,130,246,0.12); color: var(--blue-accent); } |
| .icon-cyan { background: rgba(6,182,212,0.12); color: var(--cyan); } |
| .icon-green { background: rgba(123,241,168,0.12); color: var(--green); } |
| .resource-info h4 { font-size: 15px; font-weight: 600; margin-bottom: 2px; } |
| .resource-info p { font-size: 13px; color: var(--text-muted); } |
| .resource-arrow { color: var(--text-muted); flex-shrink: 0; transition: transform 0.2s; } |
| .resource-card:hover .resource-arrow { transform: translateX(4px); color: var(--orange); } |
| |
| |
| .premium-card { |
| background: rgba(255,255,255,0.03); border: 1px solid rgba(255,255,255,0.08); |
| border-radius: 16px; padding: 24px 28px; margin-top: 20px; |
| } |
| .card-label { |
| font-size: 11px; font-weight: 700; color: var(--orange); |
| text-transform: uppercase; letter-spacing: 0.12em; margin-bottom: 8px; |
| } |
| .premium-card p { font-size: 15px; color: var(--text-secondary); line-height: 1.7; } |
| |
| |
| .stats-grid { |
| display: grid; grid-template-columns: repeat(auto-fit, minmax(260px, 1fr)); |
| gap: 16px; margin-bottom: 32px; |
| } |
| .stat-block { |
| background: var(--bg-card); border-radius: 16px; |
| padding: 28px 24px; position: relative; overflow: hidden; |
| border: 1px solid var(--border-light); |
| transition: all 0.3s; |
| } |
| .stat-block:hover { transform: translateY(-3px); box-shadow: 0 12px 32px rgba(0,0,0,0.3); } |
| .stat-block::before { |
| content: ''; position: absolute; top: 0; left: 0; right: 0; |
| height: 3px; border-radius: 3px 3px 0 0; |
| } |
| .stat-block.orange::before { background: var(--gradient-orange); } |
| .stat-block.cyan::before { background: var(--cyan); } |
| .stat-block.blue::before { background: var(--blue-accent); } |
| .stat-block.green::before { background: var(--green); } |
| .stat-value { |
| font-size: 44px; font-weight: 900; margin-bottom: 8px; |
| } |
| .stat-block.orange .stat-value { color: var(--orange); } |
| .stat-block.cyan .stat-value { color: var(--cyan); } |
| .stat-block.blue .stat-value { color: var(--blue-accent); } |
| .stat-block.green .stat-value { color: var(--green); } |
| .stat-label { font-size: 14px; color: var(--text-secondary); line-height: 1.5; } |
| |
| .metrics-row { |
| display: grid; grid-template-columns: repeat(4, 1fr); |
| background: rgba(255,255,255,0.02); border: 1px solid var(--border-subtle); |
| border-radius: 16px; overflow: hidden; |
| } |
| .metric-item { |
| padding: 20px 24px; text-align: center; |
| border-right: 1px solid var(--border-subtle); |
| } |
| .metric-item:last-child { border-right: none; } |
| .metric-val { |
| font-size: 28px; font-weight: 800; |
| background: var(--gradient-blue); |
| -webkit-background-clip: text; -webkit-text-fill-color: transparent; |
| } |
| .metric-lab { font-size: 12px; color: var(--text-muted); margin-top: 4px; text-transform: uppercase; letter-spacing: 0.06em; } |
| |
| |
| .streams-grid { |
| display: grid; grid-template-columns: repeat(3, 1fr); gap: 24px; |
| } |
| .stream-card { |
| background: var(--bg-card); border-radius: 16px; padding: 32px; |
| border: 1px solid var(--border-light); position: relative; |
| transition: all 0.3s; overflow: hidden; |
| } |
| .stream-card:hover { border-color: rgba(255,255,255,0.18); background: var(--bg-card-hover); transform: translateY(-2px); } |
| .stream-num { |
| font-size: 11px; font-weight: 700; text-transform: uppercase; |
| letter-spacing: 0.12em; margin-bottom: 16px; |
| } |
| .stream-card:nth-child(1) .stream-num { color: var(--orange); } |
| .stream-card:nth-child(2) .stream-num { color: var(--cyan); } |
| .stream-card:nth-child(3) .stream-num { color: var(--blue-accent); } |
| .stream-card h3 { font-size: 20px; font-weight: 700; margin-bottom: 8px; } |
| .stream-card .stream-count { |
| font-size: 32px; font-weight: 800; margin-bottom: 12px; |
| } |
| .stream-card:nth-child(1) .stream-count { color: var(--orange); } |
| .stream-card:nth-child(2) .stream-count { color: var(--cyan); } |
| .stream-card:nth-child(3) .stream-count { color: var(--blue-accent); } |
| .stream-card p { font-size: 14px; color: var(--text-secondary); line-height: 1.7; } |
| .stream-source { |
| display: inline-flex; align-items: center; gap: 6px; |
| font-size: 12px; font-weight: 600; color: var(--text-muted); |
| margin-top: 16px; padding: 6px 12px; border-radius: 8px; |
| background: rgba(255,255,255,0.04); border: 1px solid var(--border-subtle); |
| } |
| |
| |
| .challenge-grid { |
| display: grid; grid-template-columns: repeat(3, 1fr); gap: 24px; |
| } |
| .challenge-card { |
| background: var(--bg-card); border-radius: 16px; padding: 28px; |
| border: 1px solid var(--border-light); transition: all 0.3s; |
| } |
| .challenge-card:hover { background: var(--bg-card-hover); transform: translateY(-2px); } |
| .challenge-card h3 { font-size: 17px; font-weight: 700; margin-top: 14px; margin-bottom: 10px; } |
| .challenge-card p { font-size: 14px; color: var(--text-secondary); line-height: 1.7; } |
| |
| |
| .video-section { background: var(--bg-darker); } |
| .video-grid { |
| display: grid; grid-template-columns: 1fr 1fr; gap: 24px; |
| } |
| .video-card { |
| border-radius: 16px; overflow: hidden; |
| background: var(--bg-card); border: 1px solid var(--border-light); |
| transition: all 0.3s; |
| } |
| .video-card:hover { border-color: rgba(59,130,246,0.3); transform: translateY(-3px); box-shadow: 0 16px 48px rgba(0,0,0,0.4); } |
| .video-card.featured { |
| grid-column: 1 / -1; |
| } |
| .video-wrapper { |
| position: relative; width: 100%; aspect-ratio: 16/9; |
| background: var(--bg-dark); cursor: pointer; overflow: hidden; |
| } |
| .video-wrapper video { |
| width: 100%; height: 100%; object-fit: contain; |
| display: block; background: var(--bg-dark); |
| } |
| .video-overlay { |
| position: absolute; top: 0; left: 0; right: 0; bottom: 0; |
| display: flex; align-items: center; justify-content: center; |
| background: rgba(0,0,0,0.3); |
| transition: all 0.3s; |
| opacity: 0; pointer-events: none; |
| } |
| .video-card:hover .video-overlay { background: rgba(0,0,0,0.15); } |
| .play-btn { |
| width: 64px; height: 64px; border-radius: 50%; |
| background: rgba(255,255,255,0.15); backdrop-filter: blur(12px); |
| display: flex; align-items: center; justify-content: center; |
| transition: all 0.3s; border: 1px solid rgba(255,255,255,0.2); |
| } |
| .video-card:hover .play-btn { background: var(--orange); border-color: var(--orange); transform: scale(1.1); } |
| .play-btn svg { width: 24px; height: 24px; fill: white; margin-left: 3px; } |
| .video-info { padding: 20px 24px; } |
| .video-info h4 { font-size: 16px; font-weight: 600; margin-bottom: 6px; } |
| .video-info p { font-size: 13px; color: var(--text-muted); } |
| .video-tag { |
| display: inline-block; font-size: 11px; font-weight: 600; |
| padding: 3px 10px; border-radius: 100px; margin-bottom: 8px; |
| text-transform: uppercase; letter-spacing: 0.06em; |
| } |
| .video-tag.egocentric { background: rgba(245,166,35,0.12); color: var(--orange); } |
| .video-tag.exocentric { background: rgba(6,182,212,0.12); color: var(--cyan); } |
| .video-tag.combined { background: rgba(59,130,246,0.12); color: var(--blue-accent); } |
| |
| |
| .results-section { background: var(--bg-dark); } |
| .results-table-wrap { |
| background: var(--bg-card); border-radius: 16px; |
| border: 1px solid var(--border-light); overflow: hidden; |
| } |
| .results-table { |
| width: 100%; border-collapse: collapse; |
| } |
| .results-table thead { background: rgba(59,130,246,0.08); } |
| .results-table th { |
| padding: 16px 20px; text-align: left; |
| font-size: 12px; font-weight: 700; color: var(--text-secondary); |
| text-transform: uppercase; letter-spacing: 0.08em; |
| border-bottom: 1px solid var(--border-light); |
| } |
| .results-table td { |
| padding: 14px 20px; font-size: 14px; |
| border-bottom: 1px solid var(--border-subtle); |
| color: var(--text-secondary); |
| } |
| .results-table tr:last-child td { border-bottom: none; } |
| .results-table tr:hover td { background: rgba(255,255,255,0.02); } |
| .results-table .task-name { color: var(--text-primary); font-weight: 500; } |
| .results-table .highlight-val { color: var(--green); font-weight: 700; } |
| .results-table .baseline-val { color: var(--text-muted); } |
| .results-table .mean-row td { |
| font-weight: 700; color: var(--text-primary); |
| border-top: 2px solid var(--border-light); |
| background: rgba(245,166,35,0.04); |
| } |
| .results-table .mean-row .highlight-val { color: var(--orange); font-size: 16px; } |
| |
| |
| .pipeline-steps { |
| display: grid; grid-template-columns: repeat(4, 1fr); |
| gap: 0; position: relative; margin-top: 48px; |
| } |
| .pipeline-step { |
| text-align: center; padding: 32px 20px; position: relative; |
| } |
| .pipeline-step:not(:last-child)::after { |
| content: '→'; position: absolute; right: -8px; top: 50%; |
| transform: translateY(-50%); color: var(--orange); font-size: 20px; |
| font-weight: 700; z-index: 2; |
| } |
| .step-num { |
| width: 52px; height: 52px; border-radius: 50%; margin: 0 auto 16px; |
| display: flex; align-items: center; justify-content: center; |
| background: var(--gradient-blue); color: white; |
| font-size: 20px; font-weight: 700; |
| box-shadow: 0 4px 20px rgba(59, 130, 246, 0.3); |
| } |
| .pipeline-step h4 { font-size: 15px; font-weight: 700; margin-bottom: 8px; } |
| .pipeline-step p { font-size: 13px; color: var(--text-secondary); line-height: 1.6; } |
| |
| |
| .findings-grid { |
| display: grid; grid-template-columns: 1fr 1fr; gap: 16px; |
| } |
| .finding-card { |
| background: var(--bg-card); border: 1px solid var(--border-light); |
| border-radius: 14px; padding: 24px 28px; |
| transition: all 0.3s; |
| } |
| .finding-card:hover { border-color: rgba(245,166,35,0.2); background: var(--bg-card-hover); } |
| .finding-num { |
| font-size: 11px; font-weight: 700; color: var(--orange); |
| text-transform: uppercase; letter-spacing: 0.1em; margin-bottom: 8px; |
| } |
| .finding-card h4 { font-size: 16px; font-weight: 700; margin-bottom: 8px; } |
| .finding-card p { font-size: 14px; color: var(--text-secondary); line-height: 1.7; } |
| |
| |
| .comparison-visual { |
| display: grid; grid-template-columns: 1fr 1fr; gap: 48px; |
| margin-top: 48px; align-items: center; |
| } |
| .bar-chart { display: flex; flex-direction: column; gap: 20px; } |
| .bar-group label { |
| font-size: 13px; font-weight: 600; color: var(--text-secondary); |
| margin-bottom: 6px; display: block; |
| } |
| .bar-track { |
| width: 100%; height: 40px; background: rgba(255,255,255,0.04); |
| border-radius: 8px; position: relative; overflow: hidden; |
| } |
| .bar-fill { |
| height: 100%; border-radius: 8px; |
| display: flex; align-items: center; padding-left: 14px; |
| font-size: 14px; font-weight: 700; color: #000; |
| transition: width 1.5s cubic-bezier(0.22, 1, 0.36, 1); |
| } |
| .bar-fill.saber { background: var(--gradient-orange); } |
| .bar-fill.baseline { background: rgba(148,163,184,0.3); color: var(--text-secondary); } |
| .improvement-callout { |
| background: linear-gradient(145deg, #1c2d44, #1a2a40); |
| border: 1px solid rgba(245,166,35,0.25); border-radius: 20px; |
| padding: 40px; text-align: center; |
| } |
| .improvement-number { |
| font-size: 72px; font-weight: 900; |
| background: var(--gradient-orange); |
| -webkit-background-clip: text; -webkit-text-fill-color: transparent; |
| } |
| .improvement-label { font-size: 16px; color: var(--text-secondary); margin-top: 8px; } |
| |
| |
| .data-mix { |
| display: grid; grid-template-columns: 1fr 1fr; gap: 48px; |
| align-items: center; margin-top: 48px; |
| } |
| .donut-container { position: relative; width: 280px; height: 280px; margin: 0 auto; } |
| .donut-center { |
| position: absolute; top: 50%; left: 50%; transform: translate(-50%, -50%); |
| text-align: center; |
| } |
| .donut-center .total { font-size: 32px; font-weight: 800; color: var(--text-primary); } |
| .donut-center .total-label { font-size: 12px; color: var(--text-muted); text-transform: uppercase; letter-spacing: 0.08em; } |
| .mix-legend { display: flex; flex-direction: column; gap: 14px; } |
| .legend-item { |
| display: flex; align-items: center; gap: 12px; |
| padding: 12px 16px; border-radius: 10px; |
| background: rgba(255,255,255,0.03); border: 1px solid var(--border-subtle); |
| } |
| .legend-dot { width: 12px; height: 12px; border-radius: 3px; flex-shrink: 0; } |
| .legend-info { flex: 1; } |
| .legend-info .name { font-size: 14px; font-weight: 600; } |
| .legend-info .detail { font-size: 12px; color: var(--text-muted); } |
| .legend-pct { font-size: 16px; font-weight: 800; } |
| |
| |
| .citation-block { |
| background: rgba(255,255,255,0.03); border: 1px solid var(--border-subtle); |
| border-radius: 12px; padding: 24px; margin-top: 48px; |
| position: relative; |
| } |
| .citation-block pre { |
| font-family: 'SF Mono', 'Fira Code', monospace; |
| font-size: 13px; color: var(--text-secondary); |
| white-space: pre-wrap; line-height: 1.7; |
| } |
| .copy-btn { |
| position: absolute; top: 12px; right: 12px; |
| background: rgba(255,255,255,0.08); border: 1px solid var(--border-subtle); |
| border-radius: 8px; padding: 8px 14px; |
| font-size: 12px; font-weight: 600; color: var(--text-secondary); |
| cursor: pointer; transition: all 0.2s; |
| } |
| .copy-btn:hover { background: var(--orange); color: #000; border-color: var(--orange); } |
| |
| |
| footer { |
| background: var(--bg-darker); padding: 60px 48px 32px; |
| border-top: 1px solid var(--border-subtle); |
| } |
| .footer-inner { |
| max-width: 1200px; margin: 0 auto; |
| display: flex; justify-content: space-between; |
| } |
| .footer-brand p { font-size: 13px; color: var(--text-muted); max-width: 320px; margin-top: 12px; line-height: 1.6; } |
| .footer-links { display: flex; gap: 64px; } |
| .footer-col h4 { |
| font-size: 13px; font-weight: 600; text-transform: uppercase; |
| letter-spacing: 0.08em; margin-bottom: 16px; color: var(--text-primary); |
| } |
| .footer-col a { |
| display: block; font-size: 14px; color: var(--text-muted); |
| text-decoration: none; margin-bottom: 10px; transition: color 0.2s; |
| } |
| .footer-col a:hover { color: var(--orange); } |
| .footer-bottom { |
| max-width: 1200px; margin: 32px auto 0; |
| padding-top: 24px; border-top: 1px solid var(--border-subtle); |
| display: flex; justify-content: space-between; |
| font-size: 13px; color: var(--text-muted); |
| } |
| .footer-bottom a { color: var(--text-muted); text-decoration: none; } |
| .footer-bottom a:hover { color: var(--orange); } |
| |
| |
| .cta-banner { |
| background: linear-gradient(135deg, #1a2a4a, #0f1f3a); |
| padding: 80px 48px; text-align: center; |
| position: relative; overflow: hidden; |
| } |
| .cta-banner::before { |
| content: ''; position: absolute; top: -50%; left: -10%; width: 400px; height: 400px; |
| background: radial-gradient(circle, rgba(245,166,35,0.06) 0%, transparent 70%); |
| pointer-events: none; |
| } |
| .cta-banner::after { |
| content: ''; position: absolute; bottom: -30%; right: -5%; width: 500px; height: 500px; |
| background: radial-gradient(circle, rgba(59,130,246,0.06) 0%, transparent 70%); |
| pointer-events: none; |
| } |
| .cta-banner h2 { |
| font-size: 40px; font-weight: 800; margin-bottom: 16px; |
| position: relative; z-index: 2; |
| } |
| .cta-banner p { |
| font-size: 17px; color: var(--text-secondary); margin-bottom: 32px; |
| position: relative; z-index: 2; |
| } |
| .cta-buttons { |
| display: flex; gap: 16px; justify-content: center; |
| position: relative; z-index: 2; flex-wrap: wrap; |
| } |
| |
| |
| .fade-up { |
| opacity: 0; transform: translateY(30px); |
| transition: opacity 0.7s ease, transform 0.7s ease; |
| } |
| .fade-up.visible { opacity: 1; transform: translateY(0); } |
| |
| |
| @media (max-width: 1024px) { |
| .hero-content { grid-template-columns: 1fr; gap: 48px; } |
| .hero h1 { font-size: 44px; } |
| .streams-grid, .challenge-grid { grid-template-columns: 1fr 1fr; } |
| .pipeline-steps { grid-template-columns: 1fr 1fr; gap: 20px; } |
| .pipeline-step:not(:last-child)::after { display: none; } |
| .comparison-visual, .data-mix { grid-template-columns: 1fr; } |
| .video-grid { grid-template-columns: 1fr; } |
| .findings-grid { grid-template-columns: 1fr; } |
| nav { padding: 0 24px; } |
| section { padding: 80px 24px 60px; } |
| } |
| @media (max-width: 768px) { |
| .nav-links, .nav-buttons-desktop { display: none; } |
| .hamburger { display: flex; } |
| .hero { padding: 100px 20px 60px; min-height: auto; } |
| .hero h1 { font-size: 36px; } |
| .section-title { font-size: 28px; } |
| .streams-grid, .challenge-grid { grid-template-columns: 1fr; } |
| .pipeline-steps { grid-template-columns: 1fr; } |
| .hero-stats { grid-template-columns: 1fr; } |
| .hero-stat { border-right: none; border-bottom: 1px solid var(--border-light); } |
| .hero-stat:last-child { border-bottom: none; } |
| .metrics-row { grid-template-columns: 1fr 1fr; } |
| .metric-item:nth-child(2) { border-right: none; } |
| .metric-item:nth-child(1), .metric-item:nth-child(2) { border-bottom: 1px solid var(--border-subtle); } |
| .footer-inner { flex-direction: column; gap: 40px; } |
| .footer-links { flex-direction: column; gap: 32px; } |
| .results-table-wrap { overflow-x: auto; } |
| .cta-banner h2 { font-size: 28px; } |
| .cta-buttons { flex-direction: column; align-items: center; } |
| section { padding: 60px 16px 48px; } |
| } |
| @media (max-width: 480px) { |
| .hero h1 { font-size: 28px; } |
| .stat-value { font-size: 32px; } |
| .improvement-number { font-size: 48px; } |
| nav { height: 64px; } |
| .mobile-menu { top: 64px; } |
| } |
| </style> |
| </head> |
| <body> |
|
|
|
|
| |
| <section class="hero" id="home"> |
| <div class="hero-content"> |
| <div> |
| <div class="hero-badge"><span class="dot"></span> May 2026</div> |
| <h1>SABER<span class="highlight">.</span></h1> |
| <p class="paper-full-title">A Scalable Action-Based Embodied Dataset for Real-World VLA Adaptation — the first high-fidelity retail robotics action dataset built from natural human behavior, not teleoperation.</p> |
|
|
| <div class="premium-card" style="margin-top: 20px; margin-bottom: 28px;"> |
| <div class="card-label">The Core Claim</div> |
| <p style="font-size: 16px; font-weight: 500; line-height: 1.6; color: var(--text-primary);"> |
| Domain-specific robot deployment is fundamentally a <strong>data problem</strong>. High-fidelity naturalistic human behavior — systematically captured and retargeted — is a scalable foundation for robot adaptation. <span style="color: var(--orange);">No robot in the loop required.</span> |
| </p> |
| </div> |
|
|
|
|
| <div class="hero-stats"> |
| <div class="hero-stat"> |
| <div class="number">44.8K</div> |
| <div class="label">Training Samples</div> |
| </div> |
| <div class="hero-stat"> |
| <div class="number">100+</div> |
| <div class="label">Hours Captured</div> |
| </div> |
| <div class="hero-stat"> |
| <div class="number">2.19×</div> |
| <div class="label">Improvement</div> |
| </div> |
| </div> |
| </div> |
|
|
| |
| <div style="padding-top: 8px;"> |
| <div style="font-size: 11px; font-weight: 700; color: var(--orange); text-transform: uppercase; letter-spacing: 0.1em; margin-bottom: 14px;">Resources</div> |
| <div style="display: flex; flex-direction: column; gap: 10px;"> |
| <a href="#videos" class="resource-card" onclick="event.preventDefault();document.getElementById('videos').scrollIntoView({behavior:'smooth'});"> |
| <div class="resource-icon icon-blue"> |
| <svg width="18" height="18" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"><polygon points="23 7 16 12 23 17 23 7"/><rect x="1" y="5" width="15" height="14" rx="2" ry="2"/></svg> |
| </div> |
| <div class="resource-info"> |
| <h4>Watch Videos</h4> |
| <p>In-store capture demos</p> |
| </div> |
| <svg class="resource-arrow" width="16" height="16" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"><line x1="7" y1="17" x2="17" y2="7"/><polyline points="7 7 17 7 17 17"/></svg> |
| </a> |
|
|
| <a href="https://dreamvu.ai/saber" target="_blank" class="resource-card"> |
| <div class="resource-icon icon-orange"> |
| <svg width="18" height="18" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"><path d="M4 19.5A2.5 2.5 0 0 1 6.5 17H20"/><path d="M6.5 2H20v20H6.5A2.5 2.5 0 0 1 4 19.5v-15A2.5 2.5 0 0 1 6.5 2z"/></svg> |
| </div> |
| <div class="resource-info"> |
| <h4>arXiv</h4> |
| <p>Research Paper</p> |
| </div> |
| <svg class="resource-arrow" width="16" height="16" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"><line x1="7" y1="17" x2="17" y2="7"/><polyline points="7 7 17 7 17 17"/></svg> |
| </a> |
|
|
| <a href="DreamVu_SABER.pdf" target="_blank" class="resource-card"> |
| <div class="resource-icon icon-orange"> |
| <svg width="18" height="18" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"><path d="M21 15v4a2 2 0 0 1-2 2H5a2 2 0 0 1-2-2v-4"/><polyline points="7 10 12 15 17 10"/><line x1="12" y1="15" x2="12" y2="3"/></svg> |
| </div> |
| <div class="resource-info"> |
| <h4>Download PDF</h4> |
| <p>Paper (local copy)</p> |
| </div> |
| <svg class="resource-arrow" width="16" height="16" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"><line x1="7" y1="17" x2="17" y2="7"/><polyline points="7 7 17 7 17 17"/></svg> |
| </a> |
|
|
| <a href="https://huggingface.co/datasets/DreamVu/SABER-10K" target="_blank" class="resource-card"> |
| <div class="resource-icon icon-cyan"> |
| <svg width="18" height="18" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"><ellipse cx="12" cy="5" rx="9" ry="3"/><path d="M21 12c0 1.66-4 3-9 3s-9-1.34-9-3"/><path d="M3 5v14c0 1.66 4 3 9 3s9-1.34 9-3V5"/></svg> |
| </div> |
| <div class="resource-info"> |
| <h4>Dataset</h4> |
| <p>SABER-10K on Hugging Face</p> |
| </div> |
| <svg class="resource-arrow" width="16" height="16" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"><line x1="7" y1="17" x2="17" y2="7"/><polyline points="7 7 17 7 17 17"/></svg> |
| </a> |
|
|
| <a href="#results" class="resource-card" onclick="event.preventDefault();document.getElementById('results').scrollIntoView({behavior:'smooth'});"> |
| <div class="resource-icon icon-green"> |
| <svg width="18" height="18" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"><polyline points="23 6 13.5 15.5 8.5 10.5 1 18"/><polyline points="17 6 23 6 23 12"/></svg> |
| </div> |
| <div class="resource-info"> |
| <h4>Benchmark Results</h4> |
| <p>RoboBenchMart evaluation</p> |
| </div> |
| <svg class="resource-arrow" width="16" height="16" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"><line x1="7" y1="17" x2="17" y2="7"/><polyline points="7 7 17 7 17 17"/></svg> |
| </a> |
| </div> |
|
|
| <div style="margin-top: 20px; padding-top: 16px; border-top: 1px solid rgba(255,255,255,0.06);"> |
| <p style="color: var(--text-muted); font-size: 13px; margin-bottom: 12px;">Need the full 44.8K corpus or custom capture?</p> |
| <a href="mailto:sales@dreamvu.ai" class="btn-cta" style="font-size: 13px; padding: 8px 18px;">Contact Sales</a> |
| </div> |
| </div> |
| </div> |
| </section> |
|
|
| |
| <section style="background: var(--bg-dark); padding-top: 0; padding-bottom: 80px;"> |
| <div class="section-inner"> |
| <div class="video-card featured" style="border-radius: 20px; overflow: hidden; box-shadow: none; border: none; background: var(--bg-dark);"> |
| <div class="video-wrapper" onclick="toggleVideo(this)"> |
| <video preload="metadata" playsinline autoplay muted loop> |
| <source src="final_stitched.mp4" type="video/mp4"> |
| </video> |
| <div class="video-overlay"> |
| <div class="play-btn"> |
| <svg viewBox="0 0 24 24"><polygon points="5 3 19 12 5 21 5 3"/></svg> |
| </div> |
| </div> |
| </div> |
| <div class="video-info"> |
| <span class="video-tag combined">Full Pipeline</span> |
| <h4>Complete SABER Capture Pipeline</h4> |
| <p>The complete SABER pipeline from synchronized dual-stream videos: egocentric video, 360° exocentric view, hand landmarks, body skeleton, and SMPL mesh — derived simultaneously from real in-store human actions.</p> |
| </div> |
| </div> |
| </div> |
| </section> |
|
|
| |
| <section style="background: var(--bg-darker);"> |
| <div class="section-inner fade-up"> |
| <div class="section-label">The Challenge</div> |
| <div class="section-title">Why Retail Demands Its Own Data</div> |
| <div class="section-subtitle">Modern VLAs like GR00T N1.6 achieve near-zero success on retail tasks out of the box — not because the model is weak, but because the retail domain is entirely absent from training data.</div> |
|
|
| <div class="challenge-grid"> |
| <div class="challenge-card"> |
| <div class="resource-icon icon-orange"> |
| <svg width="20" height="20" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2"><path d="M21 16V8a2 2 0 0 0-1-1.73l-7-4a2 2 0 0 0-2 0l-7 4A2 2 0 0 0 3 8v8a2 2 0 0 0 1 1.73l7 4a2 2 0 0 0 2 0l7-4A2 2 0 0 0 21 16z"/></svg> |
| </div> |
| <h3>Distinct Skill Distribution</h3> |
| <p>Articulated object interaction, multi-height shelf reaching, basket loading, floor retrieval, and context-dependent placement — all repeated across hundreds of SKUs in layouts no lab can replicate.</p> |
| </div> |
| <div class="challenge-card"> |
| <div class="resource-icon icon-cyan"> |
| <svg width="20" height="20" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2"><path d="M1 12s4-8 11-8 11 8 11 8-4 8-11 8-11-8-11-8z"/><circle cx="12" cy="12" r="3"/></svg> |
| </div> |
| <h3>Long-Tail Scene Variation</h3> |
| <p>Dense shelves, active restocking, occlusions, varied lighting, reflective packaging, and product deformability create real-world complexity that generic datasets cannot approximate.</p> |
| </div> |
| <div class="challenge-card"> |
| <div class="resource-icon icon-blue"> |
| <svg width="20" height="20" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2"><polyline points="17 1 21 5 17 9"/><path d="M3 11V9a4 4 0 0 1 4-4h14"/><polyline points="7 23 3 19 7 15"/><path d="M21 13v2a4 4 0 0 1-4 4H3"/></svg> |
| </div> |
| <h3>Repetition Matters</h3> |
| <p>A model must see skill families repeatedly across contexts — grasping bottles from different shelf heights, opening fridges from varied approach angles — to achieve reliable deployment.</p> |
| </div> |
| </div> |
| </div> |
| </section> |
|
|
| |
| <section id="stats"> |
| <div class="section-inner fade-up"> |
| <div class="section-label">Performance</div> |
| <div class="section-title">Key Results at a Glance</div> |
|
|
| <div class="stats-grid"> |
| <div class="stat-block orange"> |
| <div class="stat-value">2.19×</div> |
| <div class="stat-label">Improvement over fine-tuning baselines on RoboBenchMart</div> |
| </div> |
| <div class="stat-block green"> |
| <div class="stat-value">29.3%</div> |
| <div class="stat-label">Mean success rate across all 10 retail manipulation tasks</div> |
| </div> |
| <div class="stat-block cyan"> |
| <div class="stat-value">91%</div> |
| <div class="stat-label">Average fridge task success — up from 43% baseline</div> |
| </div> |
| <div class="stat-block blue"> |
| <div class="stat-value">100%</div> |
| <div class="stat-label">Non-robot data — entire dataset captured from human video alone</div> |
| </div> |
| </div> |
|
|
| <div class="metrics-row"> |
| <div class="metric-item"> |
| <div class="metric-val">44.8K</div> |
| <div class="metric-lab">Total Samples</div> |
| </div> |
| <div class="metric-item"> |
| <div class="metric-val">100+</div> |
| <div class="metric-lab">Capture Hours</div> |
| </div> |
| <div class="metric-item"> |
| <div class="metric-val">3</div> |
| <div class="metric-lab">Action Streams</div> |
| </div> |
| <div class="metric-item"> |
| <div class="metric-val">10</div> |
| <div class="metric-lab">Eval Tasks</div> |
| </div> |
| </div> |
| </div> |
| </section> |
|
|
| |
| <section id="streams" style="background: var(--bg-darker);"> |
| <div class="section-inner fade-up"> |
| <div class="section-label">Dataset Architecture</div> |
| <div class="section-title">Three Complementary Action Streams</div> |
| <div class="section-subtitle">From the same dual-camera in-store captures, three distinct supervision signals are derived — each encoding a different level of kinematic abstraction.</div> |
|
|
| <div class="streams-grid"> |
| <div class="stream-card"> |
| <div class="stream-num">Stream 1</div> |
| <h3>LAPA Latent Actions</h3> |
| <div class="stream-count">25K</div> |
| <p>Embodiment-agnostic motion tokens derived via inverse-dynamics encoding from egocentric video. Captures whole-arm motion, reach trajectories, and grasping dynamics without robot joint labels.</p> |
| <div class="stream-source"> |
| <svg width="14" height="14" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2"><rect x="2" y="3" width="20" height="14" rx="2"/><line x1="8" y1="21" x2="16" y2="21"/><line x1="12" y1="17" x2="12" y2="21"/></svg> |
| Egocentric GoPro |
| </div> |
| </div> |
| <div class="stream-card"> |
| <div class="stream-num">Stream 2</div> |
| <h3>Dexterous Hand Retargets</h3> |
| <div class="stream-count">18.6K</div> |
| <p>21-point hand landmarks estimated, human-corrected frame-by-frame, then retargeted to robot joint space via Dex-Retargeting. Provides explicit finger-level precision supervision.</p> |
| <div class="stream-source"> |
| <svg width="14" height="14" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2"><rect x="2" y="3" width="20" height="14" rx="2"/><line x1="8" y1="21" x2="16" y2="21"/><line x1="12" y1="17" x2="12" y2="21"/></svg> |
| Egocentric GoPro |
| </div> |
| </div> |
| <div class="stream-card"> |
| <div class="stream-num">Stream 3</div> |
| <h3>Whole-Body Retargets</h3> |
| <div class="stream-count">1.2K</div> |
| <p>SMPL body parameters estimated from the 360° ALIA view, human-corrected, and retargeted to the Unitree G1 humanoid. Provides torso-arm-leg coordination for floor retrieval and extended reach.</p> |
| <div class="stream-source"> |
| <svg width="14" height="14" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2"><circle cx="12" cy="12" r="10"/></svg> |
| Exocentric ALIA 360° |
| </div> |
| </div> |
| </div> |
| </div> |
| </section> |
|
|
| |
| <section> |
| <div class="section-inner fade-up"> |
| <div class="section-label">Methodology</div> |
| <div class="section-title">From Store Footage to Robot Training</div> |
| <div class="section-subtitle">SABER is constructed from a dual-stream capture architecture — egocentric GoPro + exocentric ALIA 360° — across multiple real grocery stores.</div> |
|
|
| <div class="pipeline-steps"> |
| <div class="pipeline-step"> |
| <div class="step-num">1</div> |
| <h4>In-Store Capture</h4> |
| <p>100+ hours across multiple real grocery stores with head-mounted GoPro + DreamVu ALIA 360°</p> |
| </div> |
| <div class="pipeline-step"> |
| <div class="step-num">2</div> |
| <h4>Action Extraction</h4> |
| <p>LAPA encoding, hand pose estimation, and SMPL body estimation with human QC annotation</p> |
| </div> |
| <div class="pipeline-step"> |
| <div class="step-num">3</div> |
| <h4>Robot Retargeting</h4> |
| <p>Dex-Retargeting to robot hand joint space + SMPL-to-Unitree G1 whole-body retargeting</p> |
| </div> |
| <div class="pipeline-step"> |
| <div class="step-num">4</div> |
| <h4>VLA Post-Training</h4> |
| <p>Shared-backbone multi-task training on GR00T N1.6 with flow-matching objective</p> |
| </div> |
| </div> |
| </div> |
| </section> |
|
|
| |
| <section class="video-section" id="videos"> |
| <div class="section-inner fade-up"> |
| <div class="section-label">Demo Videos</div> |
| <div class="section-title">Capture Sessions & Task Annotations</div> |
| <div class="section-subtitle">Annotated in-store capture footage from the SABER dataset — showing retail manipulation tasks with action labels and multi-scene diversity.</div> |
|
|
| <div style="display: flex; flex-direction: column; gap: 32px;"> |
| |
| <div class="video-card featured" style="background: var(--bg-darker);"> |
| <div class="video-wrapper" style="background: var(--bg-darker);" onclick="toggleVideo(this)"> |
| <video preload="metadata" playsinline autoplay muted loop style="background: var(--bg-darker);"> |
| <source src="6_cycle_2.mp4" type="video/mp4"> |
| </video> |
| <div class="video-overlay"> |
| <div class="play-btn"> |
| <svg viewBox="0 0 24 24"><polygon points="5 3 19 12 5 21 5 3"/></svg> |
| </div> |
| </div> |
| </div> |
| <div class="video-info"> |
| <span class="video-tag egocentric">Annotated</span> |
| <h4>Retail Task Cycles</h4> |
| <p>Pushing trolleys, packing goods, arranging goods, opening doors, inspecting labels, and handling baskets.</p> |
| </div> |
| </div> |
|
|
| |
| <div class="video-card featured" style="background: var(--bg-darker);"> |
| <div class="video-wrapper" style="background: var(--bg-darker);" onclick="toggleVideo(this)"> |
| <video preload="metadata" playsinline autoplay muted loop style="background: var(--bg-darker);"> |
| <source src="6_cycle_1.mp4" type="video/mp4"> |
| </video> |
| <div class="video-overlay"> |
| <div class="play-btn"> |
| <svg viewBox="0 0 24 24"><polygon points="5 3 19 12 5 21 5 3"/></svg> |
| </div> |
| </div> |
| </div> |
| <div class="video-info"> |
| <span class="video-tag egocentric">Annotated</span> |
| <h4>Retail Task Cycles</h4> |
| <p>Placing and moving foods, scooping loose goods, inspecting deformable packets, carrying multiple goods, inspecting fruits, closing doors, and placing goods.</p> |
| </div> |
| </div> |
| </div> |
| </div> |
| </section> |
|
|
| |
| <section class="results-section" id="results"> |
| <div class="section-inner fade-up"> |
| <div class="section-label">Evaluation</div> |
| <div class="section-title">RoboBenchMart Results</div> |
| <div class="section-subtitle">SABER-MM post-training on GR00T N1.6 evaluated across 10 retail manipulation tasks spanning fridge, board-to-board, floor pick, and basket pick categories.</div> |
|
|
| <div class="comparison-visual"> |
| <div class="bar-chart"> |
| <div class="bar-group"> |
| <label>Mean Success — All Tasks</label> |
| <div class="bar-track"><div class="bar-fill saber" style="width: 0%;" data-width="29.3%">29.3%</div></div> |
| <div class="bar-track" style="margin-top: 6px;"><div class="bar-fill baseline" style="width: 0%;" data-width="13.4%">13.4%</div></div> |
| </div> |
| <div class="bar-group"> |
| <label>Fridge Tasks (avg open + close)</label> |
| <div class="bar-track"><div class="bar-fill saber" style="width: 0%;" data-width="91%">91%</div></div> |
| <div class="bar-track" style="margin-top: 6px;"><div class="bar-fill baseline" style="width: 0%;" data-width="43%">43%</div></div> |
| </div> |
| <div class="bar-group"> |
| <label>Floor Pick Tasks (avg)</label> |
| <div class="bar-track"><div class="bar-fill saber" style="width: 0%;" data-width="17%">17%</div></div> |
| <div class="bar-track" style="margin-top: 6px;"><div class="bar-fill baseline" style="width: 0%;" data-width="3%">3%</div></div> |
| </div> |
| </div> |
| <div class="improvement-callout"> |
| <div class="improvement-number">2.19×</div> |
| <div class="improvement-label">Mean improvement over baseline<br><span style="font-size: 13px; color: var(--text-muted);">SABER-MM vs. RoboBenchMart fine-tuning only</span></div> |
| <div style="display: flex; gap: 20px; justify-content: center; margin-top: 24px;"> |
| <div style="display: flex; align-items: center; gap: 8px;"> |
| <div style="width: 12px; height: 12px; border-radius: 3px; background: var(--gradient-orange);"></div> |
| <span style="font-size: 13px; color: var(--text-secondary);">SABER-MM</span> |
| </div> |
| <div style="display: flex; align-items: center; gap: 8px;"> |
| <div style="width: 12px; height: 12px; border-radius: 3px; background: rgba(148,163,184,0.3);"></div> |
| <span style="font-size: 13px; color: var(--text-secondary);">Baseline</span> |
| </div> |
| </div> |
| </div> |
| </div> |
|
|
| |
| <div class="results-table-wrap" style="margin-top: 48px;"> |
| <table class="results-table"> |
| <thead> |
| <tr> |
| <th>Task</th> |
| <th>Category</th> |
| <th>Baseline (RBM FT)</th> |
| <th>SABER-MM</th> |
| <th>Change</th> |
| </tr> |
| </thead> |
| <tbody> |
| <tr> |
| <td class="task-name">fridge (avg open + close)</td> |
| <td>Fridge</td> |
| <td class="baseline-val">0.43</td> |
| <td class="highlight-val">0.91</td> |
| <td style="color: var(--green);">+112%</td> |
| </tr> |
| <tr> |
| <td class="task-name">board_to_board_duff</td> |
| <td>Board</td> |
| <td class="baseline-val">0.10</td> |
| <td class="highlight-val">0.10</td> |
| <td style="color: var(--text-muted);">—</td> |
| </tr> |
| <tr> |
| <td class="task-name">board_to_board_nestle</td> |
| <td>Board</td> |
| <td class="baseline-val">0.02</td> |
| <td class="highlight-val">0.02</td> |
| <td style="color: var(--text-muted);">—</td> |
| </tr> |
| <tr> |
| <td class="task-name">board_to_board_vanish</td> |
| <td>Board</td> |
| <td class="baseline-val">0.02</td> |
| <td class="highlight-val">0.11</td> |
| <td style="color: var(--green);">+450%</td> |
| </tr> |
| <tr> |
| <td class="task-name">pick_from_floor_beans</td> |
| <td>Floor</td> |
| <td class="baseline-val">0.04</td> |
| <td class="highlight-val">0.17</td> |
| <td style="color: var(--green);">+325%</td> |
| </tr> |
| <tr> |
| <td class="task-name">pick_from_floor_slam</td> |
| <td>Floor</td> |
| <td class="baseline-val">0.02</td> |
| <td class="highlight-val">0.17</td> |
| <td style="color: var(--green);">+750%</td> |
| </tr> |
| <tr> |
| <td class="task-name">pick_to_basket_fanta</td> |
| <td>Basket</td> |
| <td class="baseline-val">0.08</td> |
| <td class="highlight-val">0.19</td> |
| <td style="color: var(--green);">+138%</td> |
| </tr> |
| <tr> |
| <td class="task-name">pick_to_basket_nivea</td> |
| <td>Basket</td> |
| <td class="baseline-val">0.08</td> |
| <td class="highlight-val">0.21</td> |
| <td style="color: var(--green);">+163%</td> |
| </tr> |
| <tr> |
| <td class="task-name">pick_to_basket_stars</td> |
| <td>Basket</td> |
| <td class="baseline-val">0.12</td> |
| <td class="highlight-val">0.14</td> |
| <td style="color: var(--green);">+17%</td> |
| </tr> |
| <tr class="mean-row"> |
| <td class="task-name">Mean (all tasks)</td> |
| <td></td> |
| <td class="baseline-val">0.134</td> |
| <td class="highlight-val">0.293</td> |
| <td style="color: var(--orange); font-weight: 800;">+119%</td> |
| </tr> |
| </tbody> |
| </table> |
| </div> |
| </div> |
| </section> |
|
|
| |
| <section style="background: var(--bg-darker);"> |
| <div class="section-inner fade-up"> |
| <div class="section-label">Training Corpus</div> |
| <div class="section-title">SABER-MM Data Composition</div> |
| <div class="section-subtitle">The post-training corpus combines SABER's three streams with robot-native anchor data and task-aligned demonstrations — totaling ~52.1K samples.</div> |
|
|
| <div class="data-mix"> |
| <div class="donut-container"> |
| <svg viewBox="0 0 200 200" width="280" height="280"> |
| |
| <circle cx="100" cy="100" r="80" fill="none" stroke="#f5a623" stroke-width="24" |
| stroke-dasharray="241 261" stroke-dashoffset="0" transform="rotate(-90 100 100)" opacity="0.9"/> |
| |
| <circle cx="100" cy="100" r="80" fill="none" stroke="#06b6d4" stroke-width="24" |
| stroke-dasharray="179 323" stroke-dashoffset="-241" transform="rotate(-90 100 100)" opacity="0.9"/> |
| |
| <circle cx="100" cy="100" r="80" fill="none" stroke="#3b82f6" stroke-width="24" |
| stroke-dasharray="46 456" stroke-dashoffset="-420" transform="rotate(-90 100 100)" opacity="0.9"/> |
| |
| <circle cx="100" cy="100" r="80" fill="none" stroke="#a855f7" stroke-width="24" |
| stroke-dasharray="24 478" stroke-dashoffset="-466" transform="rotate(-90 100 100)" opacity="0.9"/> |
| |
| <circle cx="100" cy="100" r="80" fill="none" stroke="#7BF1A8" stroke-width="24" |
| stroke-dasharray="12 490" stroke-dashoffset="-490" transform="rotate(-90 100 100)" opacity="0.9"/> |
| </svg> |
| <div class="donut-center"> |
| <div class="total">52.1K</div> |
| <div class="total-label">Total Samples</div> |
| </div> |
| </div> |
| <div class="mix-legend"> |
| <div class="legend-item"> |
| <div class="legend-dot" style="background: var(--orange);"></div> |
| <div class="legend-info"> |
| <div class="name">SABER — LAPA Latent Actions</div> |
| <div class="detail">25K samples · Egocentric video</div> |
| </div> |
| <div class="legend-pct" style="color: var(--orange);">48.0%</div> |
| </div> |
| <div class="legend-item"> |
| <div class="legend-dot" style="background: var(--cyan);"></div> |
| <div class="legend-info"> |
| <div class="name">SABER — Hand Retargets</div> |
| <div class="detail">18.6K samples · Dex-Retargeting</div> |
| </div> |
| <div class="legend-pct" style="color: var(--cyan);">35.7%</div> |
| </div> |
| <div class="legend-item"> |
| <div class="legend-dot" style="background: var(--green);"></div> |
| <div class="legend-info"> |
| <div class="name">SABER — Body Retargets</div> |
| <div class="detail">1.2K samples · Unitree G1</div> |
| </div> |
| <div class="legend-pct" style="color: var(--green);">2.3%</div> |
| </div> |
| <div class="legend-item"> |
| <div class="legend-dot" style="background: var(--blue-accent);"></div> |
| <div class="legend-info"> |
| <div class="name">NVIDIA Robot Data</div> |
| <div class="detail">4.8K samples · Anchor signal</div> |
| </div> |
| <div class="legend-pct" style="color: var(--blue-accent);">9.2%</div> |
| </div> |
| <div class="legend-item"> |
| <div class="legend-dot" style="background: #a855f7;"></div> |
| <div class="legend-info"> |
| <div class="name">RoboBenchMart</div> |
| <div class="detail">2.5K samples · Task-aligned</div> |
| </div> |
| <div class="legend-pct" style="color: #a855f7;">4.8%</div> |
| </div> |
| </div> |
| </div> |
| </div> |
| </section> |
|
|
| |
| <section id="findings"> |
| <div class="section-inner fade-up"> |
| <div class="section-label">Key Insights</div> |
| <div class="section-title">What SABER Demonstrates</div> |
|
|
| <div class="findings-grid"> |
| <div class="finding-card"> |
| <div class="finding-num">Finding 01</div> |
| <h4>Human Video Scales Where Teleoperation Can't</h4> |
| <p>SABER demonstrates that high-fidelity naturalistic human behavior, systematically captured and retargeted, is a viable and scalable foundation for domain-specific robot adaptation — without a robot in the loop.</p> |
| </div> |
| <div class="finding-card"> |
| <div class="finding-num">Finding 02</div> |
| <h4>Three Streams Are Complementary</h4> |
| <p>LAPA tokens capture whole-arm trajectory, Dex-Retargeting provides finger-level precision, and body retargets supply torso-arm-leg coordination. Together they provide non-overlapping kinematic information.</p> |
| </div> |
| <div class="finding-card"> |
| <div class="finding-num">Finding 03</div> |
| <h4>Robot-Native Anchor Stabilizes Training</h4> |
| <p>The 4,800-sample robot-native anchor data proved necessary to stabilize early training even at SABER's scale, suggesting general manipulation signal matters for robust convergence.</p> |
| </div> |
| <div class="finding-card"> |
| <div class="finding-num">Finding 04</div> |
| <h4>Task Progress Beyond Binary Success</h4> |
| <p>SABER-MM teaches models to progress further through each task sequence — mean P≥2/3 of 0.445 vs 0.278 baseline — indicating reaching and grasping are well-learned while placement remains the frontier.</p> |
| </div> |
| </div> |
| </div> |
| </section> |
|
|
| |
| <section style="background: var(--bg-darker);"> |
| <div class="section-inner fade-up"> |
| <div class="section-label">Citation</div> |
| <div class="section-title">Cite This Work</div> |
|
|
| <div class="citation-block"> |
| <button class="copy-btn" onclick="copyCitation()">Copy BibTeX</button> |
| <pre>@article{dreamvu2026saber, |
| title = {SABER: A Scalable Action-Based Embodied Dataset |
| for Real-World VLA Adaptation}, |
| author = {Menga, Narsimha and Sakurikar, Parikshit and Rouhi, Amirreza |
| and Reddy, Satya Sai and Govil, Anirudh and Chittajallu, Sri Harsha |
| and Aggarwal, Rajat and Namboodiri, Anoop and Reddi, Sashi}, |
| year = {2026}, |
| month = {May}, |
| note = {DreamVu Inc.}, |
| url = {https://dreamvu.ai/saber} |
| }</pre> |
| </div> |
| </div> |
| </section> |
|
|
| |
| <section class="cta-banner"> |
| <h2>Ready to Build the <span style="background: var(--gradient-orange); -webkit-background-clip: text; -webkit-text-fill-color: transparent;">Data Layer</span> for Retail Robots?</h2> |
| <p>The SABER-10K subset is available now. Full dataset and code at dreamvu.ai/saber.</p> |
| <div class="cta-buttons"> |
| <a href="https://huggingface.co/datasets/DreamVu/SABER-10K" target="_blank" class="btn-cta" style="font-size: 16px; padding: 14px 32px;">Download SABER-10K on HuggingFace</a> |
| <a href="https://dreamvu.ai/saber" target="_blank" class="btn-outline" style="font-size: 16px; padding: 14px 32px;">Full Paper & Dataset →</a> |
| <a href="mailto:sales@dreamvu.ai" class="btn-outline" style="font-size: 16px; padding: 14px 32px;">Contact Sales</a> |
| </div> |
| </section> |
|
|
|
|
| <script> |
| |
| function toggleMenu() { |
| document.getElementById('mobileMenu').classList.toggle('open'); |
| document.querySelector('.hamburger').classList.toggle('active'); |
| } |
| function closeMenu() { |
| document.getElementById('mobileMenu').classList.remove('open'); |
| document.querySelector('.hamburger').classList.remove('active'); |
| } |
| |
| |
| function toggleVideo(wrapper) { |
| const video = wrapper.querySelector('video'); |
| const overlay = wrapper.querySelector('.video-overlay'); |
| if (video.paused) { |
| |
| document.querySelectorAll('.video-wrapper video').forEach(v => { |
| if (v !== video) { v.pause(); v.closest('.video-wrapper').querySelector('.video-overlay').style.opacity = '1'; } |
| }); |
| video.play(); |
| overlay.style.opacity = '0'; |
| } else { |
| video.pause(); |
| overlay.style.opacity = '1'; |
| } |
| } |
| |
| |
| function copyCitation() { |
| const text = document.querySelector('.citation-block pre').textContent; |
| navigator.clipboard.writeText(text).then(() => { |
| const btn = document.querySelector('.copy-btn'); |
| btn.textContent = 'Copied!'; |
| setTimeout(() => btn.textContent = 'Copy BibTeX', 2000); |
| }); |
| } |
| |
| |
| const observer = new IntersectionObserver((entries) => { |
| entries.forEach(entry => { |
| if (entry.isIntersecting) { |
| entry.target.classList.add('visible'); |
| |
| entry.target.querySelectorAll('.bar-fill').forEach(bar => { |
| setTimeout(() => { bar.style.width = bar.dataset.width; }, 300); |
| }); |
| } |
| }); |
| }, { threshold: 0.15 }); |
| |
| document.querySelectorAll('.fade-up').forEach(el => observer.observe(el)); |
| |
| </script> |
|
|
| </body> |
| </html> |
| |
