Driving · 2026-10-01
ReWAM: Reciprocal World Action Models for Interactive Autonomous Driving
Benshan Ma, Pei Liu, Ruiguo Zhong, Lang Zhang, Mingyue Feng, Yaonong Wang, Jun Ma
AI-assisted summary based on the paper. Consult the linked source for methods and limitations. Benchmark numbers appear after editor verification.
TL;DR
ReWAM is a game-theoretic world action modeling framework for interactive autonomous driving that explicitly models reciprocal influence between the ego agent and other agents. It uses a Level-k response hierarchy with role-specific ego/other Action DiTs that exchange strategy tokens via cross-agent attention, grounded in a shared future-world latent representation. The model is learned from demonstrations using conditional flow matching.
Why it matters
The authors argue that existing World Action Models condition on ego actions but treat other agents as part of the predicted world rather than strategic responders, which can fail in dense interaction. ReWAM introduces bounded reciprocal reasoning inside a generative world-action model, and they report state-of-the-art results on NAVSIM with larger gains in critical interaction scenarios.
Method
- ReWAM represents ego and other agents as mutually conditioned responders using a finite Level-k hierarchy to avoid solving a joint equilibrium.
- A shared latent future-world representation from observations and navigation condition grounds role-specific Ego/Other Action DiTs; cross-agent attention conditions each level on preceding-level strategy hypotheses.
- Expert actions are treated as samples from the best-response distribution; the hierarchy is trained by joint conditional flow matching (distribution matching) over demonstration support.
Limitation
The paper states that Level-k avoids computing a fully rational multi-agent equilibrium by using a finite hierarchy of bounded responses.
Abstract (from arXiv)
In interactive scenarios, an autonomous driving system is required to generate ego actions under the influence of other agents' behaviors. Existing World Action Models (WAMs) typically model other agents as components of the world model rather than as decision-makers that fundamentally shape the action of the ego agent, which impairs their performance in dense interaction scenarios. We introduce Reciprocal World Action Models (ReWAM), a game-theoretic world action modeling framework that captures the reciprocal influence between the ego agent and other agents by representing them as conditional responders whose actions are mutually influenced. We instantiate this framework with a Level-$k$ response hierarchy, where role-specific ego and other action DiTs exchange compact strategy tokens through cross-agent attention while remaining grounded in a shared representation of the future driving world. To learn the response policy of the ego agent from demonstrations, we formulate expert actions as samples from the best response distribution and jointly optimize the entire hierarchy using conditional flow matching. Our framework is evaluated on the NAVSIM dataset and achieves state-of-the-art performance compared to baselines. The improvement is particularly significant in interactive scenarios, validating that modeling reciprocal responses provides a more effective foundation for interaction-aware world action generation.
Related papers
- Multi-Agent Egocentric World Model with Fine-Grained Embodied Interaction
- UniW(A)M: A Unified Mixture-of-Experts Framework for World and World-Action Models
- PhysWAM: Physically Consistent World Action Model for Autonomous Driving
- OccWAM: Occupancy World-Action Models with Joint Scene-to-Ego Flow Matching
- V2X-WAM: A Cooperative World Action Model for End-to-End Autonomous Driving
Publication record
Bibliographic evidence describes where and when a record appeared. It does not validate scientific claims.
First public release: 2026-09-30 · Indexed by Pub-AI: 2026-10-01 UTC
ReWAM: Reciprocal World Action Models for Interactive Autonomous Driving
Preprint posted: 2026-09-30 · Original record ↗
- arxiv
- 2609.39245
Explicit source versions (1)
- v1 · 2026-09-30
- Preprint available
Source date: 2026-09-30 · First detected: 2026-10-25 UTC · Historical baseline
Source evidence (1)
- Official evidence ↗ · atom:published (official metadata)
Abstract · arxiv
In interactive scenarios, an autonomous driving system is required to generate ego actions under the influence of other agents' behaviors. Existing World Action Models (WAMs) typically model other agents as components of the world model rather than as decision-makers that fundamentally shape the action of the ego agent, which impairs their performance in dense interaction scenarios. We introduce Reciprocal World Action Models (ReWAM), a game-theoretic world action modeling framework that captures the reciprocal influence between the ego agent and other agents by representing them as conditional responders whose actions are mutually influenced. We instantiate this framework with a Level-$k$ response hierarchy, where role-specific ego and other action DiTs exchange compact strategy tokens through cross-agent attention while remaining grounded in a shared representation of the future driving world. To learn the response policy of the ego agent from demonstrations, we formulate expert actions as samples from the best response distribution and jointly optimize the entire hierarchy using conditional flow matching. Our framework is evaluated on the NAVSIM dataset and achieves state-of-the-art performance compared to baselines. The improvement is particularly significant in interactive scenarios, validating that modeling reciprocal responses provides a more effective foundation for interaction-aware world action generation.