# CRL-VLA: Continual Vision-Language-Action Learning

arXiv: https://arxiv.org/abs/2602.03445 · HTML: https://mlsanity.com/paper/2602.03445/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- Placed in Reinforcement learning and post-training, one of 13 directions in Vision-language-action models.
- Builds on 32 of its 39 references that are in the map.
- First posted on arXiv 3 Feb 2026.

## Placed in

- Vision-language-action models / [Reinforcement learning and post-training](https://mlsanity.com/vla/reinforcement-learning-and-post-training/)

## Builds on (most cited first, 30 of 32)

- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [Learning without Forgetting](https://mlsanity.com/paper/1606.09282/) (2016)
- [Gradient Episodic Memory for Continual Learning](https://mlsanity.com/paper/1706.08840/) (2017)
- [π0.5: a Vision-Language-Action Model with Open-World Generalization](https://mlsanity.com/paper/2504.16054/) (2025)
- [LIBERO: Benchmarking Knowledge Transfer for Lifelong Robot Learning](https://mlsanity.com/paper/2306.03310/) (2023)
- [Experience Replay for Continual Learning](https://mlsanity.com/paper/1811.11682/) (2018)
- [Constrained Policy Optimization](https://mlsanity.com/paper/1705.10528/) (2017)
- [Fine-Tuning Vision-Language-Action Models: Optimizing Speed and Success](https://mlsanity.com/paper/2502.19645/) (2025)
- [Progress & Compress: A scalable framework for continual learning](https://mlsanity.com/paper/1805.06370/) (2018)
- [Policy Distillation](https://mlsanity.com/paper/1511.06295/) (2015)
- [TinyVLA: Towards Fast, Data-Efficient Vision-Language-Action Models for Robotic Manipulation](https://mlsanity.com/paper/2409.12514/) (2024)
- [Reinforced Continual Learning](https://mlsanity.com/paper/1805.12369/) (2018)
- [RL's Razor: Why Online Reinforcement Learning Forgets Less](https://mlsanity.com/paper/2509.04259/) (2025)
- [SimpleVLA-RL: Scaling VLA Training via Reinforcement Learning](https://mlsanity.com/paper/2509.09674/) (2025)
- [RL with KL penalties is better viewed as Bayesian inference](https://mlsanity.com/paper/2205.11275/) (2022)
- [Continual World: A Robotic Benchmark For Continual Reinforcement Learning](https://mlsanity.com/paper/2105.10919/) (2021)
- [Improving Vision-Language-Action Model with Online Reinforcement Learning](https://mlsanity.com/paper/2501.16664/) (2025)
- [ChatVLA: Unified Multimodal Understanding and Robot Control with Vision-Language-Action Model](https://mlsanity.com/paper/2502.14420/) (2025)
- [Actions as Language: Fine-Tuning VLMs into VLAs Without Catastrophic Forgetting](https://mlsanity.com/paper/2509.22195/) (2025)
- [ReinboT: Amplifying Robot Visual-Language Manipulation with Reinforcement Learning](https://mlsanity.com/paper/2505.07395/) (2025)
- [Same State, Different Task: Continual Reinforcement Learning without Interference](https://mlsanity.com/paper/2106.02940/) (2021)
- [IRL-VLA: Training an Vision-Language-Action Policy via Reward World Model](https://mlsanity.com/paper/2508.06571/) (2025)
- [RLinf-VLA: A Unified and Efficient Framework for Reinforcement Learning of Vision-Language-Action Models](https://mlsanity.com/paper/2510.06710/) (2025)
- [Elastic Weight Consolidation (EWC): Nuts and Bolts](https://mlsanity.com/paper/2105.04093/) (2021)
- [Robust Finetuning of Vision-Language-Action Robot Policies via Parameter Merging](https://mlsanity.com/paper/2512.08333/) (2025)
- [A Survey of Continual Reinforcement Learning](https://mlsanity.com/paper/2506.21872/) (2025)
- [Balancing Signal and Variance: Adaptive Offline RL Post-Training for VLA Flow Models](https://mlsanity.com/paper/2509.04063/) (2025)
- [Revisiting Replay and Gradient Alignment for Continual Pre-Training of Large Language Models](https://mlsanity.com/paper/2508.01908/) (2025)
- [Continually Evolving Skill Knowledge in Vision Language Action Model](https://mlsanity.com/paper/2511.18085/) (2025)
- [Stabilizing Elastic Weight Consolidation method in practical ML tasks and using weight importances for neural network pruning](https://mlsanity.com/paper/2109.10021/) (2021)

## Built on by (most cited first, 7 of 7)

- [CORAL: Scalable Multi-Task Robot Learning via LoRA Experts](https://mlsanity.com/paper/2603.09298/) (2026)
- [Information-Theoretic Constraints for Continual Vision-Language-Action Alignment](https://mlsanity.com/paper/2603.13335/) (2026)
- [FAN: Foresight Action Normalization for Continual Adaptation of Vision-Language-Action Models](https://mlsanity.com/paper/2609.21358/) (2026)
- [SAMBAR: Selective Anchoring via Method of Multipliers for Balanced Knowledge Acquisition and Retention in Vision-Language-Action Models](https://mlsanity.com/paper/2609.32108/) (2026)
- [Train Together or Merge Later? Unifying VLA Experts via a Shared Action Interface](https://mlsanity.com/paper/2609.33125/) (2026)
- [RIFAR: Reliability and Forgetting-Aware Replay for Continual Robot Learning](https://mlsanity.com/paper/2610.03079/) (2026)
- [ACCRUE: Continual Off-Policy Reinforcement Learning for Robot Foundation Models](https://mlsanity.com/paper/2610.18596/) (2026)

## Abstract (arXiv)

Lifelong learning is critical for embodied agents in open-world environments, where reinforcement learning fine-tuning has emerged as an important paradigm to enable Vision-Language-Action (VLA) models to master dexterous manipulation through environmental interaction. Thus, Continual Reinforcement Learning (CRL) is a promising pathway for deploying VLA models in lifelong robotic scenarios, yet balancing stability (retaining old skills) and plasticity (learning new ones) remains a formidable challenge for existing methods. We introduce CRL-VLA, a framework for continual post-training of VLA models with rigorous theoretical bounds. We derive a unified performance bound linking the stability-plasticity trade-off to goal-conditioned advantage magnitude, scaled by policy divergence. CRL-VLA resolves this dilemma via asymmetric regulation: constraining advantage magnitudes on prior tasks while enabling controlled growth on new tasks. This is realized through a simple but effective dual-critic architecture with novel Goal-Conditioned Value Formulation (GCVF), where a frozen critic anchors semantic consistency and a trainable estimator drives adaptation. Experiments on the LIBERO benchmark demonstrate that CRL-VLA effectively harmonizes these conflicting objectives, outperforming baselines in both anti-forgetting and forward adaptation.