New MAPS architecture improves autonomous vehicle coordination at intersections

Researchers have developed a new hierarchical deep reinforcement learning architecture called the Master-Agent Proto-plan System (MAPS) to address the challenge of coordinating autonomous vehicles at unsignalized intersections. MAPS uses a central Master agent to generate a compact embedding, or proto-plan, that guides decentralized Worker agents in executing vehicle control. This approach successfully demonstrated collision-free navigation and reduced travel times in simulations, with a system trained on three agents showing robust generalization to five-agent scenarios. AI

IMPACT This research could lead to more efficient and safer coordination for autonomous vehicle fleets in complex urban environments.

RANK_REASON Academic paper detailing a new system for multi-agent reinforcement learning. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.MA (Multiagent) →

• arXiv
• Autonomous Vehicles
• Deep Reinforcement Learning
• HighwayEnv
• Master Agent
• Master-Agent Proto-plan System
• Multi-agent Reinforcement Learning
• Unsignalized Intersections
• Worker Agents

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →