Compact Latent Coordination for Autonomous Vehicles at Unsignalized Intersections

Source

• Source name: arXiv
• Source kind: paper
• Authors / organization: Gil Lifshits, Igal Bilik, Gilad Katz
• Venue: arXiv: cs.LG
• Published: 2026-07-23
• Collected: 2026-07-24
• URL: http://arxiv.org/abs/2607.21488v1
• PDF: https://arxiv.org/pdf/2607.21488v1
• DOI: UNKNOWN

Plain-Language Brief

Researchers present MAPS, a deep reinforcement learning architecture for coordinating autonomous vehicles at unsignalized intersections. The system uses a central Master agent that generates a global coordination strategy and decentralized Worker agents for vehicle-specific control. Tests showed that MAPS outperformed state-of-the-art baselines in collision-free navigation and reducing average travel time.

What the Document Says

• The paper proposes MAPS, a hierarchical deep reinforcement learning architecture.
• MAPS consists of a centralized Master agent and decentralized Worker agents.
• The Master agent generates a compact, continuous embedding (proto-plan) that encodes a global coordination strategy.
• Worker agents integrate the proto-plan with local observations to execute vehicle-specific control.
• The system is tested across 72 intersection configurations in HighwayEnv.

Key Concepts Mentioned

• autonomous vehicles
• unsignalized intersections
• multi-agent reinforcement learning
• hierarchical deep reinforcement learning
• MAPS
• proto-plan
• Master agent
• Worker agents
• collision-free navigation
• average travel time

Claims Made by the Document

• MAPS outperforms state-of-the-art baselines in collision-free navigation and reducing average travel time.

Evidence Used

• Tests across 72 intersection configurations in HighwayEnv showed that MAPS achieves collision-free navigation while significantly reducing average travel time, outperforming state-of-the-art baselines.

Methods / Approach

UNKNOWN

Limitations Stated by the Authors

UNKNOWN

Limitations Noted by the Agent

Agent interpretation — not from the source.

UNKNOWN

Open Questions

UNKNOWN

Notes for Future Q&A

May include interpretation. Labeled where it does.

UNKNOWN

Raw Abstract / Excerpt

Coordinating autonomous vehicles at unsignalized intersections remains a critical challenge for multi-agent reinforcement learning (MARL) systems, which typically struggle with combinatorial action spaces, reliance on privileged information, or rigid agent designs. We propose Master-Agent Proto-plan System (MAPS), a hierarchical deep reinforcement learning (DRL) architecture in which a centralized Master agent generates a compact, continuous embedding, denoted as proto-plan, that encodes a global coordination strategy. Decentralized Worker agents integrate this embedding with local observations to execute vehicle-specific control, decoupling strategic intent from tactical execution and enabling independent optimization of each module. As a proof-of-concept evaluation of this coordination mechanism, we test MAPS across 72 intersection configurations in HighwayEnv. MAPS achieves collision-free navigation while significantly reducing average travel time, outperforming state-of-the-art baselines. The learned proto-plans further exhibit robust generalization: a system trained with three agents achieves a 94% success rate when deployed zero-shot to five-agent scenarios, confirming that proto-plan-based hierarchical learning provides a promising framework for multi-vehicle coordination.