Overview
Why join
Grand Challenge 2026 brings together benchmark tracks in reasoning, multimodal perception, cooperative intelligence, robustness, and real-world autonomy.
Tracks share data releases and evaluation tooling with DriveX workshops and research assets, making participation accessible for both new and established teams.
Definition
About the Grand Challenge
The DriveX Grand Challenge is the benchmarking pillar of the broader DriveX initiative. It unifies multiple challenge tracks under a shared research vision for physical AI and intelligent transportation.
Rather than operating as disconnected competitions, tracks are organized under one framework for community engagement, measurable progress, and technical exchange across multimodal perception, cooperative intelligence, robustness, and reasoning.
In addition to competition outcomes, the program is designed to inform long-term research priorities across the DriveX ecosystem.
Rationale
Why the DriveX Grand Challenge
Progress in autonomy and intelligent transportation requires open, structured, and reproducible evaluation in realistic conditions. Key advances are needed in perception, prediction, control, reasoning, generalization, and deployment robustness.
The DriveX Grand Challenge organizes these directions within a single initiative framework, supporting technical depth within each track while improving cross-track integration.
This approach also provides continuity across years and venues, allowing the challenge program to mature as a long-term research platform.
Thematic structure
2026 challenge themes
For the 2026 cycle, the DriveX Grand Challenge highlights tracks spanning three broad themes.
Cooperative and multimodal perception
Multimodal scene understanding, sensor fusion, infrastructure-assisted perception, and cooperative intelligence across vehicles, roadside systems, and heterogeneous sensing environments.
Robustness, generalization, and heterogeneity
Domain generalization, heterogeneous data conditions, cross-domain robustness, and model performance under realistic shifts in environment, sensing, and operational context.
Reasoning and high-level scene intelligence
High-level scene understanding, embodied reasoning, and advanced semantic and contextual intelligence for real-world autonomous systems.
Tracks
Challenge tracks
The following tracks are part of the DriveX Grand Challenge 2026. Each card opens the official challenge or benchmark site in a new tab; the DriveX overview link stays on this website.
Preview images use a placeholder until you drop screenshots under images/captures/ (same basename as suggested in each card’s alt text is optional).
Cooperative and multimodal perception
MDrive Track
Robustness, generalization, and heterogeneity
doScenes Track
HetroD Track
Reasoning and high-level scene intelligence
Reasoning Track
How the Grand Challenge fits within DriveX
The Grand Challenge is one of DriveX’s core recurring programs.
DriveX provides shared scientific direction, initiative identity, and long-term continuity across workshops, challenges, research assets, and community activities. Within this framework, the Grand Challenge coordinates multiple tracks and organizing teams under one benchmark-driven umbrella.
This structure keeps the program open and collaborative while maintaining long-term coherence across cycles.
Timeline
The DriveX Grand Challenge 2026 runs throughout the 2026 cycle, including track announcements, data releases, submission milestones, and associated workshop activities.
Please refer to the individual track pages for specific timelines, data access details, evaluation procedures, and participation instructions.
Organizers
The DriveX Grand Challenge is coordinated by the DriveX team with support from track organizers and collaborators across technical domains.
Organizers — initiative contacts and partner coordination are summarized on the People page; each track lists its own technical organizers.
Track organizers — see individual track pages for full organizer listings.
Resources
Track-specific datasets, benchmark descriptions, evaluation protocols, submission instructions, and related technical resources are available through the individual challenge pages and associated DriveX Research Assets.
As the broader DriveX ecosystem continues to grow, the Grand Challenge will also help connect benchmark activities with workshops, datasets, and related open technical resources.
Looking ahead
The DriveX Grand Challenge is designed to expand over time through broader technical coverage, stronger continuity across tracks, and deeper links to workshops and research assets.
By combining open participation with a shared initiative vision, the program helps advance physical AI and intelligent transportation in a structured, collaborative format.
The DriveX Grand Challenge 2026 is part of the broader DriveX initiative led by UCLA and developed with partner institutions and collaborators.