WAM-AD @ WACV 2027

About the Workshop

The first WAM-AD workshop will convene researchers in computer vision, autonomous driving, generative modeling, and machine learning who are working toward vehicles that can both predict how a traffic scene changes under intervention and produce executable driving actions. World Action Models (WAMs) couple action-conditioned world modeling with policy learning, bridging generative video, 3D/4D scene understanding, motion forecasting, planning, and Vision-Language-Action (VLA) models.

WAM-AD will be a half-day, in-person, single-track workshop at WACV 2027 (Disney Springs, Lake Buena Vista, FL). A single-track program will combine invited talks, presentations of all accepted papers, structured Q&A, and a closing community discussion.

We welcome algorithms, systems, datasets, evaluations, and critical analyses on topics including but not limited to:

Action-conditioned traffic-scene prediction, counterfactual simulation, and generative driving world models
Cascaded and joint WAM architectures; VLA models for perception, prediction, planning, and control
3D/4D scene, occupancy, map, and agent representations; dynamics and traffic-rule reasoning
Model-based planning, reinforcement/imitation learning, memory, and long-horizon closed-loop driving
Scalable data engines, digital twins, synthetic corner cases, and simulation-to-reality transfer
Driving datasets, benchmarks, and metrics for visual fidelity, physical consistency, safety, and task success
Uncertainty, causal reasoning, interpretability, robustness, human-vehicle interaction, and oversight
Efficient on-vehicle WAMs, connected/cooperative driving, and transfer across vehicle platforms

The preliminary schedule is available here.

Call for Papers

Accepted papers will appear in the WACV 2027 Workshop Proceedings. Submissions will follow the WACV format and undergo double-blind peer review through OpenReview. We will accept archival regular papers of 5–8 pages (including references). Every accepted paper will receive a 10-minute oral presentation slot, including Q&A.

Submit papers on OpenReview: https://openreview.net/group?id=thecvf.com/WACV/2027/Workshop/WAM-AD. Please refer to the Call for Papers for more details.

Call for Speakers

We invite researchers and industry experts to share their work on world action models, autonomous driving, and related topics as invited speakers at WAM-AD. Please submit your information via the Call for Speakers form.

Call for Reviewers

We welcome volunteers to join our program committee and help review workshop submissions. If you are interested, please sign up through the Call for Reviewers form.