AnchorVLA: Bridging Discrete Decisions and Continuous Trajectories for Vision-Language-Action Planning

2026-07-03

AI Summary

Existing vision-language-action (VLA) planning approaches in autonomous driving struggle to align high-level semantic decisions with continuous trajectory generation, suffering from discretization errors, inefficient inference, and misalignment between semantics and actions. This work proposes a hierarchical decision-anchored VLA planning framework that introduces a novel “decision-as-anchor” representation and an anchored residual flow mechanism. By explicitly linking high-level semantic decisions to low-level execution through trajectory-mode anchors, the method generates fine-grained continuous trajectories within the residual space defined by these anchors, enabling efficient decoupling and precise alignment between semantics and actions. Integrated with large language model–driven decision reasoning and multimodal fusion, the approach achieves a 77.28% success rate and an 89.92 driving score on the Bench2Drive closed-loop benchmark, significantly outperforming current state-of-the-art methods.

Abstract

Autonomous driving planning requires translating navigation intent, traffic rules, dynamic interactions, and language instructions into executable continuous trajectories. Vision-Language-Action models have been introduced into driving planning to improve long-tail generalization, commonsense reasoning, high-level semantic understanding, and explainability. However, existing VLA planners mainly follow planning-head-based trajectory prediction or full-trajectory autoregressive generation. The former only weakly constrains continuous trajectory generation with VLA reasoning, while the latter relies on long sequences of low-information-density coordinate tokens, making semantic-action alignment difficult and leading to discretization errors and inefficient inference. To address these limitations, we propose AnchorVLA, a hierarchical decision-anchored VLA planning framework that uses trajectory-pattern anchors as an explicit interface between high-level VLA reasoning and continuous trajectory execution. Specifically, Decision-as-Anchor Representation represents behavior-level driving decisions with anchor tokens, each encoding an entire local motion pattern rather than a single coordinate point. Decision-Anchored Residual Flow then generates fine-grained continuous trajectories in the selected anchor-defined residual space, capturing multi-modal execution refinements after high-level decision making. By reasoning over compact and semantically meaningful anchors instead of autoregressively generating waypoint sequences, AnchorVLA preserves LLM-based decision making while improving inference efficiency, semantic-action alignment, and continuous generation flexibility. Experiments on the Bench2Drive closed-loop benchmark show that AnchorVLA achieves a state-of-the-art Success Rate of 77.28 and a competitive Driving Score of 89.92.

Problem

• Vision-Language-Action
• autonomous driving planning
• continuous trajectory generation
• semantic-action alignment
• discretization errors

Innovation

• Vision-Language-Action
• trajectory anchors
• hierarchical planning
• decision-anchored representation
• continuous trajectory generation