Efficient Real-World Online Reinforcement Learning for Robot Manipulation via Centralized Training and Critic Decomposition

• URL: http://arxiv.org/abs/2608.09762v1
• Date: Mon, 10 Aug 2026 15:54:25 GMT
• Title: Efficient Real-World Online Reinforcement Learning for Robot Manipulation via Centralized Training and Critic Decomposition
• Abstract summary: We introduce a unified framework that combines centralized training with decentralized execution (CTDE) and a Hybrid Reward Architecture (HRA)<n> Experimental results demonstrate that the proposed framework substantially improves both sample efficiency and policy performance.<n>Compared with a state-of-the-art baseline, our method improves the success rate from 60% to 80% on tennis ball pick-and-place.
• Score: 16.480150894507908
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Real-world online reinforcement learning (RL) provides a promising approach for training robotic manipulation policies directly in the physical world, avoiding the sim-to-real gap and enabling continuous policy refinement through human-in-the-loop interaction. Recent methods have demonstrated sample-efficient learning through human intervention but remain limited to small randomization ranges and encounter challenges with the non-stationarity induced by concurrently training multiple agents. To address these limitations, we introduce a unified framework that combines centralized training with decentralized execution (CTDE) and a Hybrid Reward Architecture (HRA). This enables multiple actors to share a centralized multi-head critic. The critic is decomposed into task and grasp heads, corresponding to the sparse task reward and a potential-based grasping reward, respectively. We accordingly reformulate the critic and actor objectives to exploit the decomposed Q-values while explicitly accounting for the categorical action distribution of the discrete gripper policy. Experimental results demonstrate that the proposed framework substantially improves both sample efficiency and policy performance. We validate our approach on two robotic arms and a simulated humanoid robot across tennis ball and banana pick-and-place, pot reset, and simulated block relocation tasks under dimension-wise domain randomization, approximately 5-25x larger than those considered in prior work. Compared with a state-of-the-art baseline, our method improves the success rate from 60% to 80% on tennis ball pick-and-place, from 60% to 90% on banana pick-and-place, and from 25% to 95% on simulated block relocation, while also successfully accomplishing a task where the baseline consistently fails. Videos and more details are available at our project website: https://hil-harc.github.io/.

Related papers

• Max-Q Selective Imitation for Human-in-the-Loop Online Robot Learning [1.96243636752331]
  Human-in-the-loop (HIL) online reinforcement learning for real robots must absorb human interventions quickly while continuing to improve beyond the human prior.<n>We present a training method for this setting based on two components.<n>emphMC Q-chunk critic regresses chunk-level action values onto Monte Carlo returns from a replay buffer.<n>emphmax-Q selective imitation updates the actor by imitating, at each state, the higher-$Q$ action between the current policy action and a buffer sample.
  arXiv  Detail & Related papers  (2026-08-15T07:13:59Z)
• A Rubric-Supervised Critic from Sparse Real-World Outcomes [87.11204512676193]
  Real-world coding agents operate with humans in the loop, where success signals are typically noisy, delayed, and sparse.<n>We propose a process to learn a "critic" model from sparse and noisy interaction data, which can then be used both as a reward model for either RL-based training or inference-time scaling.
  arXiv  Detail & Related papers  (2026-03-04T07:23:54Z)
• Actor-Critic Pretraining for Proximal Policy Optimization [2.7575165233051693]
  This paper proposes a pretraining approach for actor-critic algorithms like Proximal Policy Optimization (PPO)<n>The actor is pretrained via behavioral cloning, while the critic is pretrained using returns obtained from rollouts of the pretrained policy.<n> Experimental results show that actor-critic pretraining improves sample efficiency by 86.1% on average compared to no pretraining and by 30.9% to actor-only pretraining.
  arXiv  Detail & Related papers  (2026-02-27T08:43:56Z)
• Flow Policy Gradients for Robot Control [67.61978635211048]
  Flow matching policy gradients can be made effective for training and fine-tuning more expressive policies.<n>We show how policies can exploit the flow representation for exploration when training from scratch, as well as improved fine-tuning robustness over baselines.
  arXiv  Detail & Related papers  (2026-02-02T18:56:49Z)
• Residual Off-Policy RL for Finetuning Behavior Cloning Policies [41.99435186991878]
  We present a recipe that combines the benefits of behavior cloning (BC) and reinforcement learning (RL) through a residual learning framework.<n>Our method requires only sparse binary reward signals and can effectively improve manipulation policies on high-degree-of-freedom (DoF) systems.<n>In particular, we demonstrate, to the best of our knowledge, the first successful real-world RL training on a humanoid robot with dexterous hands.
  arXiv  Detail & Related papers  (2025-09-23T17:59:46Z)
• FLaRe: Achieving Masterful and Adaptive Robot Policies with Large-Scale Reinforcement Learning Fine-Tuning [74.25049012472502]
  FLaRe is a large-scale Reinforcement Learning framework that integrates robust pre-trained representations, large-scale training, and gradient stabilization techniques. Our method aligns pre-trained policies towards task completion, achieving state-of-the-art (SoTA) performance on previously demonstrated and on entirely novel tasks and embodiments.
  arXiv  Detail & Related papers  (2024-09-25T03:15:17Z)
• REBEL: Reward Regularization-Based Approach for Robotic Reinforcement Learning from Human Feedback [61.54791065013767]
  A misalignment between the reward function and human preferences can lead to catastrophic outcomes in the real world.<n>Recent methods aim to mitigate misalignment by learning reward functions from human preferences.<n>We propose a novel concept of reward regularization within the robotic RLHF framework.
  arXiv  Detail & Related papers  (2023-12-22T04:56:37Z)
• Robot Fine-Tuning Made Easy: Pre-Training Rewards and Policies for Autonomous Real-World Reinforcement Learning [58.3994826169858]
  We introduce RoboFuME, a reset-free fine-tuning system for robotic reinforcement learning. Our insights are to utilize offline reinforcement learning techniques to ensure efficient online fine-tuning of a pre-trained policy. Our method can incorporate data from an existing robot dataset and improve on a target task within as little as 3 hours of autonomous real-world experience.
  arXiv  Detail & Related papers  (2023-10-23T17:50:08Z)
• Reinforcement Learning with Foundation Priors: Let the Embodied Agent Efficiently Learn on Its Own [59.11934130045106]
  We propose Reinforcement Learning with Foundation Priors (RLFP) to utilize guidance and feedback from policy, value, and success-reward foundation models. Within this framework, we introduce the Foundation-guided Actor-Critic (FAC) algorithm, which enables embodied agents to explore more efficiently with automatic reward functions. Our method achieves remarkable performances in various manipulation tasks on both real robots and in simulation.
  arXiv  Detail & Related papers  (2023-10-04T07:56:42Z)
• Obstacle Avoidance for Robotic Manipulator in Joint Space via Improved Proximal Policy Optimization [6.067589886362815]
  In this paper, we train a deep neural network via an improved Proximal Policy Optimization (PPO) algorithm to map from task space to joint space for a 6-DoF manipulator. Since training such a task in real-robot is time-consuming and strenuous, we develop a simulation environment to train the model. Experimental results showed that using our method, the robot was capable of tracking a single target or reaching multiple targets in unstructured environments.
  arXiv  Detail & Related papers  (2022-10-03T10:21:57Z)
• Robot Learning of Mobile Manipulation with Reachability Behavior Priors [38.49783454634775]
  Mobile Manipulation (MM) systems are ideal candidates for taking up the role of a personal assistant in unstructured real-world environments. Among other challenges, MM requires effective coordination of the robot's embodiments for executing tasks that require both mobility and manipulation. We study the integration of robotic reachability priors in actor-critic RL methods for accelerating the learning of MM for reaching and fetching tasks.
  arXiv  Detail & Related papers  (2022-03-08T12:44:42Z)
• Reactive Long Horizon Task Execution via Visual Skill and Precondition Models [59.76233967614774]
  We describe an approach for sim-to-real training that can accomplish unseen robotic tasks using models learned in simulation to ground components of a simple task planner. We show an increase in success rate from 91.6% to 98% in simulation and from 10% to 80% success rate in the real-world as compared with naive baselines.
  arXiv  Detail & Related papers  (2020-11-17T15:24:01Z)

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.