Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Thinking While Driving: A Concurrent Framework for Real-Time, LLM-Based Adaptive Routing

Xiaopei Tan, Muyang Fan

TL;DR

<3-5 sentence high-level summary>

Abstract

We present Thinking While Driving, a concurrent routing framework that integrates LLMs into a graph-based traffic environment. Unlike approaches that require agents to stop and deliberate, our system enables LLM-based route planning while agents are moving, significantly reducing intersection wait times. Under high traffic, agents average just 0.75 seconds of decision latency. To coordinate many agents in real-time, we implement a non-blocking asynchronous architecture using Unity coroutines and a dedicated request manager. The environment is a weighted undirected graph with live congestion metrics, updated continuously by the agents to enable shared perception. Our results show LLM-driven agents can dynamically adapt to traffic, reroute around congestion, and exhibit behaviors beyond static pathfinding, all while maintaining real-time performance. This work provides a reproducible framework for future research in adaptive routing and multi-agent cooperation.

Thinking While Driving: A Concurrent Framework for Real-Time, LLM-Based Adaptive Routing

TL;DR

<3-5 sentence high-level summary>

Abstract

We present Thinking While Driving, a concurrent routing framework that integrates LLMs into a graph-based traffic environment. Unlike approaches that require agents to stop and deliberate, our system enables LLM-based route planning while agents are moving, significantly reducing intersection wait times. Under high traffic, agents average just 0.75 seconds of decision latency. To coordinate many agents in real-time, we implement a non-blocking asynchronous architecture using Unity coroutines and a dedicated request manager. The environment is a weighted undirected graph with live congestion metrics, updated continuously by the agents to enable shared perception. Our results show LLM-driven agents can dynamically adapt to traffic, reroute around congestion, and exhibit behaviors beyond static pathfinding, all while maintaining real-time performance. This work provides a reproducible framework for future research in adaptive routing and multi-agent cooperation.

Paper Structure

This paper contains 35 sections, 4 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Agent-Based Traffic and Driving Models
  4. LLM Agents for Transportation Decision-Making
  5. Systems for Low-Latency LLM Inference
  6. Contribution: The Identified Gap
  7. Methodology
  8. System Architecture
  9. Component Implementation
  10. Agent Control (TrafficAI)
  11. Lifecycle Management (Spawner)
  12. Global State Tracking (RoadManager)
  13. Asynchronous Inference Bridge (LLMManager)
  14. Environment Modeling and Shared Perception
  15. Dual-Representation of Cost
  16. ...and 20 more sections

Figures (4)

  • Figure 1: Overview of the concurrent traffic simulation architecture. Agents interact with the physical environment while asynchronously querying the Cognitive Layer for routing decisions.
  • Figure 2: Comparison of Sequential (Stop-and-Think) vs. Concurrent (Thinking While Driving) reasoning models. The concurrent model overlaps inference time with travel time.
  • Figure 3: Impact of the Thinking While Driving framework on Intersection Wait Times. The concurrent architecture effectively eliminates the reasoning delay.
  • Figure 4: Average Journey Time comparison across Low and High Density scenarios. LLM agents outperform A* in high-density settings due to adaptive rerouting.