Closed-Loop Action Chunks with Dynamic Corrections for Training-Free Diffusion Policy

Pengyuan Wu; Pingrui Zhang; Zhigang Wang; Dong Wang; Bin Zhao; Xuelong Li

cs.ROMar 2, 2026

Closed-Loop Action Chunks with Dynamic Corrections for Training-Free Diffusion Policy

Authors: Pengyuan Wu, Pingrui Zhang, Zhigang Wang, Dong Wang, Bin Zhao, Xuelong Li

Organizations: Zhejiang University. · Shanghai Artificial Intelligence Laboratory. · Fudan University. · Northwestern Polytechnical University. ·

Abstract

Diffusion-based policies have achieved remarkable results in robotic manipulation but often struggle to adapt rapidly in dynamic scenarios, leading to delayed responses or task failures. We present DCDP, a Dynamic Closed-Loop Diffusion Policy framework that integrates chunk-based action generation with real-time correction. DCDP integrates a self-supervised dynamic feature encoder, cross-attention fusion, and an asymmetric action encoder-decoder to inject environmental dynamics before action execution, achieving real-time closed-loop action correction and enhancing the system's adaptability in dynamic scenarios. In dynamic PushT simulations, DCDP improves adaptability by 19% without retraining while requiring only 5% additional computation. Its modular design enables plug-and-play integration, achieving both temporal coherence and real-time responsiveness in dynamic robotic scenarios, including real-world manipulation tasks. The project page is at: https://github.com/wupengyuan/dcdp

Figures & tables

Fig. 1: Comparison of the Open-Loop Diffusion Policy, Closed-Loop Diffusion Policy, and Our Closed-Loop with Dynamic Correction Diffusion Policy. The orange line depicts the action chunking prediction of length $H$ for the Open-Loop Diffusion Policy. Meanwhile, the green line denotes the Closed-Loop Diffusion Policy, which employs single-step inference to achieve closed-loop control. However, this approach incurs high latency and requires frequent re-planning. By contrast, our Dynamic Correction Closed-Loop Diffusion Policy (as indicated by the red line ) leverages information from a length- $M$ History Bank to perform lightweight, fast corrections at each inference step, thereby achieving closed-loop control.

Fig. 2: Overview of the DCDP. Our method adopts a two-stage framework. In Stage 1 (left panel of the figure), we train the Fast Dynamic-Aware Policy and a variational autoencoder (VAE). In Stage 2 (right panel of the figure), we apply training-free, per-step action corrections using the aforementioned Fast Dynamic-Aware Policy; the corrected actions are then decoded by the VAE decoder.

Fig. 3: Different types of perturbations are considered, where the random perturbation updates its direction at certain time steps.

Methods	Static	Constant Perturbations	Random Perturbations
Open-Loop	88.4	58.2	52.8
Close-Loop	84.6	76.1	61.6
Temporal Ensemble	81.0	65.8	57.3
DCDP	92.5	77.6	71.9

TABLE I: Success rates across three levels of perturbation intensity show that the proposed DCDP method outperforms all baseline methods.

Methods	OL(H=8)	CL(H=1)	TE(H=1)	DCDP(H=8)
Delay (ms)	7.05	53.60	53.74	7.39

TABLE II: Comparison of per-step inference latency. DCDP adds only 5% overhead in closed-loop execution, substantially lower than other closed-loop methods.

Fig. 4: Visualization of how various inference strategies respond to perturbations.

TA	SSD	DCA	SR ${}_{\text{static}}$	SR ${}_{\text{dynamic}}$
			88.40	58.20
	✓	✓	92.05	73.59
✓		✓	92.31	73.50
✓	✓		93.36	68.23
✓	✓	✓	92.50	77.60

TABLE III: Task success rates after ablating individual modules, demonstrating that each module in DCDP is effective.

Fig. 5: The two tasks comprise two types of perturbations: constant-direction and random-direction. These perturbations were applied exclusively to the components highlighted in the figure.

Diffusion Policy

Diffusion policies generate multimodal action sequences by iteratively denoising noise, enabling imitation, planning, and control in robotics.

See all 205 papers

Set-Supervised Diffusion Policy: Learning Action-Chunking Diffusion through Corrections
Jun 1, 2026Zhaoting Li, Gang Chen, Javier Alonso-Mora +2Diffusion Policy Robotic Manipulation
Delft University of Technology · University of Stuttgart
AdaWorldPolicy: World-Model-Driven Diffusion Policy with Online Adaptive Learning for Robotic Manipulation
Feb 23, 2026Ge Yuan, Qiyuan Qiao, Jing Zhang +1Robot Policy Adaptation Robotic Manipulation
The University of Hong Kong · Beihang University
Dynamic Neural Koopman Distillation for Fast Robot Control Using Diffusion Models
May 24, 2026Lei Zheng, Peiqi Yu, Zengqi Peng +2Robotic Control Koopman Operator Learning
Department of Electrical and Computer Engineering, National University of Singapore, Singapore · Department of Electrical and Computer Engineering, Carnegie Mellon University, USA · Robotics and Autonomous Systems Thrust, The Hong Kong University of Science and Technology, Hong Kong SAR, China

Jun 1, 2026J/K move · Enter open · S save

Set-Supervised Diffusion Policy: Learning Action-Chunking Diffusion through Corrections

Zhaoting Li, Gang Chen, Javier Alonso-Mora, Cosimo Della Santina+1

Delft University of Technology · University of Stuttgart

Diffusion Policy Robotic Manipulation Learning from Demonstration Robot Imitation Learning

Open paper AskPDF arXiv

Diffusion policies have recently emerged as a powerful framework for robotic manipulation. However, like other behavior cloning methods, they remain vulnerable to distributional shift, often requiring human-in-the-loop interventions to correct failures during deployment. These interactions naturally provide paired supervision in the form of the robot's undesired actions and the human teacher's corrective actions. Yet existing data aggregation pipelines and standard behavior cloning losses largely ignore this negative signal from undesired actions, leading to overfitting to teacher's actions and an increasing reliance on costly expert data. To address this limitation, we propose Set-Supervised Diffusion Policy (SDP), a novel learning framework that utilizes contrastive action-chunk data to train diffusion policies from human corrections. From paired positive and negative action-chunks, SDP constructs a set of desired action-chunks and designs a training pipeline that encourages the diffusion policy to align with the set. Through extensive experiments across multiple robotic manipulation tasks, we demonstrate that SDP consistently improves policy performance, with particularly strong gains in robustness to noisy data. Moreover, SDP induces high-quality aggregated datasets, enabling more efficient and reliable policy learning from human-in-the-loop corrections. Our code is available at https://set-supervised-diffusion-policy.github.io/.

Abstract

Diffusion Policy

Explore similar work

Set-Supervised Diffusion Policy: Learning Action-Chunking Diffusion through Corrections

AdaWorldPolicy: World-Model-Driven Diffusion Policy with Online Adaptive Learning for Robotic Manipulation

Dynamic Neural Koopman Distillation for Fast Robot Control Using Diffusion Models