mlsanityData to 7 Oct 2026

RouteRLT: Learning When and Which RL Specialist Should Control a Vision-Language-Action Policy

arXiv 2609.26467 ↗First posted 22 Sep 2026cs.ROcited by 0

On the map: Vision-language-action models → Reinforcement learning and post-training

Abstract (arXiv)

Vision-language-action (VLA) models often struggle in the precision-critical phases of multi-stage manipulation tasks. To mitigate this issue, VLA models can be used in conjunction with reinforcement learning (RL) specialists that are specifically trained to handle the precision-critical phases. However, the coordination between the base VLA model and the RL specialists, which dictates when a specialist should take over from the base VLA and vice-versa, remains an open research question. In this paper, we address this gap by introducing RouteRLT, a modular framework that coordinates a generalist VLA, used as the default controller, with designated precision-critical RL specialists. At a high level, our framework trains a phase-aware coordination mechanism that handles handoffs between the generalist and the specialists. We evaluate RouteRLT on the LIBERO and LIBERO-Plus benchmarks, as well as on a physical connector pickup and insertion task. Overall, we find that RouteRLT improves success on LIBERO, and retains net gains on LIBERO-Plus. On the physical task, RouteRLT completes 65.7% of trials, compared with 8.6% for the baseline. Altogether, these results demonstrate that learned coordination builds on generalist VLA capabilities to improve task completion in precision-critical manipulation.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Reinforcement learning and post-training, one of 13 directions in Vision-language-action models. Builds on 32 of its 36 references that are in the map. First posted on arXiv 22 Sep 2026.

Builds on

32 of 36 references.

Closest ancestors · 5

Same direction, newest first

  1. 2026Foresight Residual RL for Long-Horizon Robot Manipulation with Vision-Language-Action Models17 Jul
  2. 2026RL Token: Bootstrapping Online RL with Vision-Language-Action Models24 Apr
  3. 2025STARE-VLA: Progressive Stage-Aware Reinforcement for Fine-Tuning Vision-Language-Action Models4 Dec
  4. 2025ConRFT: A Reinforced Fine-tuning Method for VLA Models via Consistency Policy8 Feb
  5. 2024Steering Your Generalists: Improving Robotic Foundation Models via Value Guidance17 Oct

Other references · 27

  1. 2024OpenVLA: An Open-Source Vision-Language-Action Modelcited by 2,777
  2. 2018Addressing Function Approximation Error in Actor-Critic Methodscited by 2,302
  3. 2024π0: A Vision-Language-Action Flow Model for General Robot Controlcited by 2,297
  4. 2023LIBERO: Benchmarking Knowledge Transfer for Lifelong Robot Learningcited by 1,338
  5. 2025Fine-Tuning Vision-Language-Action Models: Optimizing Speed and Successcited by 844
  6. 2025SmolVLA: A Vision-Language-Action Model for Affordable and Efficient Roboticscited by 507
  7. 2025X-VLA: Soft-Prompted Transformer as Scalable Cross-Embodiment Vision-Language-Action Modelcited by 285
  8. 2020Recovery RL: Safe Reinforcement Learning with Learned Recovery Zonescited by 156
  9. 2024Precise and Dexterous Robotic Manipulation via Human-in-the-Loop Reinforcement Learningcited by 153
  10. 2025Steering Your Diffusion Policy with Latent Space Reinforcement Learningcited by 149
  11. 2025VLA-Adapter: An Effective Paradigm for Tiny-Scale Vision-Language-Action Modelcited by 140
  12. 2026VLA-JEPA: Enhancing Vision-Language-Action Model with Latent World Modelcited by 104
15 more
  1. 2022Robot Learning on the Job: Human-in-the-Loop Autonomy and Learning During Deploymentcited by 101
  2. 2024Unpacking Failure Modes of Generative Policies: Runtime Monitoring of Consistency and Progresscited by 63
  3. 2024Sparse Diffusion Policy: A Sparse, Reusable, and Flexible Policy for Robot Learningcited by 56
  4. 2021LazyDAgger: Reducing Context Switching in Interactive Imitation Learningcited by 50
  5. 2024RLDG: Robotic Generalist Policy Distillation via Reinforcement Learningcited by 42
  6. 2026AtomicVLA: Unlocking the Potential of Atomic Skill Learning in Robotscited by 15
  7. 2026PACE: Phase-Aware Chunk Execution for Robot Policies with Action Chunkingcited by 7
  8. 2025MoIRA: Modular Instruction Routing Architecture for Multi-Task Roboticscited by 4
  9. 2026Continue or Replan? Bernoulli-Continuation Policy Learning for Adaptive Horizon Executioncited by 3
  10. 2025MuST: Multi-Head Skill Transformer for Long-Horizon Dexterous Manipulation with Skill Progresscited by 2
  11. 2026RPG: Robust Policy Gating for Smooth Multi-Skill Transitions in Humanoid Fightingcited by 2
  12. 2026TRACT: Temporally Routed Action Chunks with Chronological Phase Authority for Contact-Rich Manipulationcited by 2
  13. 2026WireCraft: A Simulation Benchmark for Industrial DLO Manipulationcited by 1
  14. 2026PAMAE: Phase-Aware-MoE Action Experts Towards Reliable Flow-Matching Vision-Language-Action Policiescited by 1
  15. 2026Think Only When Needed: Prompt-Authority Control for Selective Slow-Path Intervention in Vision-Language-Action Manipulationcited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026SWAP: Stepwise Action Policy Routing for Vision-Language-Action Models
  2. 2026Vision-Language-Action Jump-Starting for Reinforcement Learning Robotic Agents
  3. 2026IG-RFT: An Interaction-Guided RL Framework for VLA Models in Long-Horizon Robotic Manipulation
  4. 2025VLA-RL: Towards Masterful and General Robotic Manipulation with Scalable Reinforcement Learning
  5. 2026Z-1: Efficient Reinforcement Learning for Vision-Language-Action Models
  6. 2026EXPO-FT: Sample-Efficient Reinforcement Learning Finetuning for Vision-Language-Action Models
  7. 2026eRLT: Efficient VLA Reinforcement Learning via Action-Relevant Token Routing
  8. 2026Find Something You Can't Do: Agentic Real-World Reinforcement Learning for Self-Improving VLA Models