| <!DOCTYPE html> |
| <html lang="en"> |
| <head> |
| <meta charset="UTF-8" /> |
| <meta name="viewport" content="width=device-width, initial-scale=1.0" /> |
| <title>AV Causal Scenario Retrieval Challenge</title> |
| <link rel="stylesheet" href="style.css" /> |
| </head> |
| <body> |
|
|
| <nav class="top"> |
| <div class="inner"> |
| <div class="brand">AV Causal Scenario<span class="dot">.</span> Retrieval Challenge</div> |
| <ul> |
| <li><a href="#overview">Overview</a></li> |
| <li><a href="#track">Track</a></li> |
| <li><a href="#timeline">Timeline</a></li> |
| <li><a href="#organizers">Organizers</a></li> |
| </ul> |
| </div> |
| </nav> |
|
|
| <header class="hero"> |
| <div class="hero-banner"> |
| <img src="assets/banner.png" alt="AV Causal Scenario Retrieval Challenge banner" /> |
| </div> |
| <div class="hero-content"> |
| <div class="eyebrow">2026 · NVIDIA <a href="https://research.nvidia.com/labs/sil/" target="_blank">Spatial Intelligence Lab</a></div> |
| <h1 class="hero-title">AV Causal Scenario Retrieval Challenge</h1> |
| <p class="hero-lead"> |
| Retrieve complex driving scenarios from dense spatio-temporal-causal annotations of real-world videos. |
| </p> |
| </div> |
| </header> |
|
|
| <main> |
|
|
| |
| <section id="overview"> |
| <h2><span class="sec-num">01</span> Overview</h2> |
| <p class="section-lead"> |
| Understanding complex driving scenarios requires more than object detection. |
| It demands reasoning about the causal chains that connect agents, actions, |
| and environmental context over time. The AV Causal Scenario Retrieval |
| Challenge invites the research community to develop models that can retrieve |
| driving scenarios from dense, structured annotations capturing spatial, |
| temporal, and causal relationships. |
| </p> |
| <p class="section-lead"> |
| The challenge is built on the CASCADE dataset, |
| which provides dense human annotations for NVIDIA's public |
| <a href="https://huggingface.co/datasets/nvidia/PhysicalAI-Autonomous-Vehicles" target="_blank">PAI-AV Dataset</a>. |
| CASCADE captures hierarchical, temporally dense relationships across real-world |
| dashcam video including multi-entity-class schemas, explicit causal links, |
| spatial containment hierarchies, and normative compliance flags. |
| </p> |
| </section> |
|
|
| |
| <section id="track"> |
| <h2><span class="sec-num">02</span> Challenge Track</h2> |
| <p class="section-lead"> |
| Details on challenge tracks, submission format, evaluation protocol, and |
| scoring will be announced soon. |
| </p> |
| <div class="card"> |
| <span class="pill">Track · Retrieval</span> |
| <h3>Causal Scenario Retrieval</h3> |
| <p> |
| <b>Task:</b> Given a natural-language query describing a complex driving |
| situation involving spatial, temporal, and/or causal relationships, |
| retrieve the matching scenario(s) from the annotated corpus. |
| </p> |
| <p class="muted"> |
| Full task specification and baseline details will be published soon. |
| </p> |
| </div> |
| </section> |
|
|
| |
| <section id="timeline"> |
| <h2><span class="sec-num">03</span> Timeline</h2> |
| <ol class="timeline"> |
| <li> |
| <span class="date">2026-06-03</span> |
| <b>Challenge announcement</b> |
| </li> |
| <li class="future"> |
| <span class="date" aria-label="Opening rescheduled from July 31, 2026 to August 14, 2026"><del><time datetime="2026-07-31">2026-07-31</time></del> → <strong><time datetime="2026-08-14">2026-08-14</time></strong></span> |
| <b>Competition goes live</b> |
| <span class="desc">Registration, public data, starter tools, and submission instructions open.</span> |
| </li> |
| <li class="future"> |
| <span class="date">2026-11-20</span> |
| <b>Public leaderboard closes</b> |
| <span class="desc">Final submissions accepted for the leaderboard.</span> |
| </li> |
| <li class="future"> |
| <span class="date">2026-11-29</span> |
| <b>Final results released</b> |
| <span class="desc">Final results and winners are released to participants.</span> |
| </li> |
| <li class="future"> |
| <span class="date">NeurIPS 2026</span> |
| <b>Public winner announcement at NeurIPS 2026</b> |
| <span class="desc">Top entries presented at a workshop session.</span> |
| </li> |
| </ol> |
| <p class="muted">Dates are tentative and subject to update.</p> |
| </section> |
|
|
| |
| <section id="organizers"> |
| <h2><span class="sec-num">04</span> Organizers</h2> |
| <p class="section-lead"> |
| This challenge is organized by NVIDIA's Spatial Intelligence Lab. |
| </p> |
| <a class="card organizer-card" href="https://research.nvidia.com/labs/sil/" target="_blank" rel="noopener noreferrer"> |
| <span class="pill">Host</span> |
| <h3>NVIDIA Spatial Intelligence Lab</h3> |
| <p>NVIDIA Research lab working on spatial intelligence for autonomous systems, 3D understanding, and scene reasoning.</p> |
| </a> |
| </section> |
|
|
| </main> |
|
|
| <footer> |
| © 2026 NVIDIA · Spatial Intelligence Lab |
| </footer> |
|
|
| </body> |
| </html> |
|
|
