FutureX: Enhance End-to-End Autonomous Driving via Latent Chain-of-Thought World Model

2025-12-11

AI Summary

To address the insufficient robustness of end-to-end planners in dynamic traffic—stemming from their neglect of ego-vehicle–environment interaction—this paper proposes a world model framework grounded in implicit Chain-of-Thought (CoT) reasoning. Our method introduces an Auto-think Switch mechanism that adaptively alternates between “thinking mode” (multi-step CoT rollouts in latent space) and “reactive mode” (single-step response), enabling both future scene evolution modeling and progressive trajectory optimization. We pioneer the integration of implicit CoT into end-to-end driving planning, synergizing lightweight scene summarization with a Latent World Model to balance reasoning depth and real-time performance. Evaluated on NAVSIM, our approach significantly improves TransFuser’s PDMS by +6.2 while reducing collision rates, all while maintaining millisecond-level inference latency—thus achieving a favorable trade-off between safety and computational efficiency.

Technology Category

• Planning, Routing, and Scheduling: Model-Based Reasoning
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Multiagent Systems: Multiagent Planning

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

In autonomous driving, end-to-end planners learn scene representations from raw sensor data and utilize them to generate a motion plan or control actions. However, exclusive reliance on the current scene for motion planning may result in suboptimal responses in highly dynamic traffic environments where ego actions further alter the future scene. To model the evolution of future scenes, we leverage the World Model to represent how the ego vehicle and its environment interact and change over time, which entails complex reasoning. The Chain of Thought (CoT) offers a promising solution by forecasting a sequence of future thoughts that subsequently guide trajectory refinement. In this paper, we propose FutureX, a CoT-driven pipeline that enhances end-to-end planners to perform complex motion planning via future scene latent reasoning and trajectory refinement. Specifically, the Auto-think Switch examines the current scene and decides whether additional reasoning is required to yield a higher-quality motion plan. Once FutureX enters the Thinking mode, the Latent World Model conducts a CoT-guided rollout to predict future scene representation, enabling the Summarizer Module to further refine the motion plan. Otherwise, FutureX operates in an Instant mode to generate motion plans in a forward pass for relatively simple scenes. Extensive experiments demonstrate that FutureX enhances existing methods by producing more rational motion plans and fewer collisions without compromising efficiency, thereby achieving substantial overall performance gains, e.g., 6.2 PDMS improvement for TransFuser on NAVSIM. Code will be released.

Problem

• Enhance autonomous driving planners via future scene reasoning
• Model ego-environment interaction with Chain-of-Thought world models
• Improve motion planning quality and reduce collisions dynamically

Innovation

• Latent World Model predicts future scene representations
• Auto-think Switch decides when to use reasoning
• Chain-of-Thought guides trajectory refinement for planning