Multi-agent LLM architectures are transforming autonomous driving by addressing key challenges in single-agent systems.

• Enhanced perception capabilities through specialized agent collaboration
• Improved decision-making via distributed intelligence and task specialization
• Reduced computational demands by distributing processing across multiple agents
• Higher safety standards through redundancy and collaborative verification

This research matters for engineering by establishing frameworks for more robust autonomous systems that better handle complex driving scenarios with lower computational overhead.

Multi-Agent Autonomous Driving Systems with Large Language Models: A Survey of Recent Advances