Latent-Centroid Steering: Single-Pass Classifier-Free Guidance for Command-Aligned Autonomous Driving

• URL: http://arxiv.org/abs/2608.00237v1
• Date: Fri, 31 Jul 2026 19:32:28 GMT
• Title: Latent-Centroid Steering: Single-Pass Classifier-Free Guidance for Command-Aligned Autonomous Driving
• Abstract summary: We propose a single-pass guidance mechanism that replaces instance-level residuals with class-level latent shifts.<n>We demonstrate that LCS reduces inference latency by approximately 50% while achieving stronger command adherence.
• Score: 37.533498955258544
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Vision-language models (VLMs) have recently emerged as a promising paradigm for end-to-end autonomous driving, enabling agents to map multimodal inputs and high-level navigation instructions directly to executable trajectories. However, in practice, these models exhibit a persistent command-following gap: predicted trajectories often show weak sensitivity to navigation commands, resulting in incorrect behavior at critical decision points. We identify this issue as a form of conditional policy collapse, where regression-based training under multimodal trajectory distributions encourages the model to rely on dominant visual priors while marginalizing the language-conditioned signal. To address this issue, we introduce a principled formulation of classifier-free guidance (CFG) for regression-based vision-language driving. We show that CFG can be interpreted as isolating the instruction-induced residual in the action space by contrasting conditional and unconditional predictions, thereby explicitly amplifying the effect of the navigation command at inference time. However, a standard two-pass CFG introduces prohibitive latency for real-time control and produces noisy instance-level guidance directions. Building on a mean-shift interpretation of CFG, we propose Latent-Centroid Steering (LCS), a single-pass guidance mechanism that replaces instance-level residuals with class-level latent shifts. By projecting conditional representations toward precomputed command-specific centroids, LCS performs class-level latent steering based on cluster geometry that is both more stable and computationally efficient. We demonstrate that LCS reduces inference latency by approximately 50% while achieving stronger command adherence and improved driving performance on both closed-loop (Bench2Drive) and open-loop (nuScenes) benchmarks. Code will be released.

Related papers

• End-to-end Conditional Diffusion for Realistic and Controllable Visual Traffic Scenario Generation [59.188006170485046]
  E2E-CDiff is an end-to-end conditional diffusion framework for controllable and realistic scenario generation.<n>Differentiable guidance regulates speed, enforces drivable-area compliance, and supports collision-avoidance or collision-seeking behaviors.<n>E2E-CDiff also performs competitively as a learning-based ego planner.
  arXiv  Detail & Related papers  (2026-07-21T02:13:25Z)
• AnchorVLA: Bridging Discrete Decisions and Continuous Trajectories for Vision-Language-Action Planning [13.949868529278504]
  AnchorVLA is a hierarchical decision-anchored VLA planning framework.<n>It uses trajectory-pattern anchors as an explicit interface between high-level VLA reasoning and continuous trajectory execution.<n>It achieves a state-of-the-art Success Rate of 77.28 and a competitive Driving Score of 89.92.
  arXiv  Detail & Related papers  (2026-07-03T10:38:30Z)
• CLEAR: Cognition and Latent Evaluation for Adaptive Routing in End-to-End Autonomous Driving [19.474428775260034]
  We propose CLEAR, a framework that combines ultra-fast generative planning with deep semantic reasoning.<n> CLEAR employs Drive-JEPA as the visual encoder and replaces the multi-step denoising chain with a single-step conditional drift in a VAE latent space.<n>On the NAVSIM v1 benchmark, CLEAR achieves a state-of-the-art PDMS of 93.7.
  arXiv  Detail & Related papers  (2026-06-04T14:32:10Z)
• Closed-Loop Neural Activation Control in Vision-Language-Action Models [19.22723477431829]
  We propose a closed-loop framework that replaces static intervention strength with adaptive, time-varying control signals.<n>We instantiate this framework with both PID and reinforcement learning based controllers.<n> Experiments show that residual-STEER achieves more stable concept regulation and a better steering-task success trade-off than fixed-coefficient baselines.
  arXiv  Detail & Related papers  (2026-05-29T18:59:53Z)
• OneVL: One-Step Latent Reasoning and Planning with Vision-Language Explanation [61.18260993245354]
  Chain-of-Thought (CoT) reasoning is a powerful driver of trajectory prediction in VLA-based autonomous driving.<n>We present OneVL, a unified VLA and World Model framework that routes reasoning through compact latent tokens supervised by dual auxiliary decoders.<n>OneVL becomes the first latent CoT method to surpass explicit CoT, delivering state-of-the-art accuracy at answer-only latency.
  arXiv  Detail & Related papers  (2026-04-20T16:37:22Z)
• Open-Ended Instruction Realization with LLM-Enabled Multi-Planner Scheduling in Autonomous Vehicles [40.195332742819865]
  This study proposes an instruction-realization framework that leverages a large language model (LLM) to interpret instructions.<n>It generates executable scripts that schedule multiple model predictive control (MPC)-based motion planners based on real-time feedback, and converts planned trajectories into control signals.
  arXiv  Detail & Related papers  (2026-04-09T09:32:21Z)
• ICR-Drive: Instruction Counterfactual Robustness for End-to-End Language-Driven Autonomous Driving [1.6961250145588112]
  We introduce ICR-Drive, a diagnostic framework for instruction counterfactual robustness in language-conditioned autonomous driving.<n>We replay identical CARLA routes under matched simulator configurations and seeds to isolate performance changes attributable to instruction language.<n>Experiments on LMDrive and BEVDriver show that minor instruction changes can induce substantial performance drops and distinct failure modes.
  arXiv  Detail & Related papers  (2026-04-07T03:24:22Z)
• Why Steering Works: Toward a Unified View of Language Model Parameter Dynamics [81.80010043113445]
  Local weight fine-tuning, LoRA-based adaptation, and activation-based interventions are studied in isolation.<n>We present a unified view that frames these interventions as dynamic weight updates induced by a control signal.<n>Across methods, we observe a consistent trade-off between preference and utility: stronger control increases preference while predictably reducing utility.
  arXiv  Detail & Related papers  (2026-02-02T17:04:36Z)
• GrAInS: Gradient-based Attribution for Inference-Time Steering of LLMs and VLMs [56.93583799109029]
  GrAInS is an inference-time steering approach that operates across both language-only and vision-language models and tasks.<n>During inference, GrAInS hidden activations at transformer layers guided by token-level attribution signals, and normalizes activations to preserve representational scale.<n>It consistently outperforms both fine-tuning and existing steering baselines.
  arXiv  Detail & Related papers  (2025-07-24T02:34:13Z)
• KV Cache Steering for Controlling Frozen LLMs [80.50365534625438]
  cache steering is a lightweight method for implicit steering of language models.<n>We apply cache steering to induce chain-of-thought reasoning in small language models.
  arXiv  Detail & Related papers  (2025-07-11T17:59:36Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.