aiqbench analysis
This research introduces MemCorr-DP, a diffusion policy that enhances robotic manipulation by explicitly modeling 3D geometric relations between current scenes and reference trajectories. By employing counterfactual conditioning, the method significantly improves closed-loop success rates when objects move or camera viewpoints shift beyond standard training distributions. The work advances machine learning in robotics by demonstrating that explicit spatial correspondence is more robust than relying solely on future motion or centroid geometry for generalization.
MemCorr-DP: Counterfactual Correspondence Conditioning for a Diffusion Policy Guided by a Reference
arXiv:2609.06615v1 Announce Type: new Abstract: Behavior-cloned visuomotor policies can remain accurate near their training distribution yet fail when object position and camera viewpoint change together. A successful reference trajectory contains the geometry needed to transfer the same interaction, but the policy must align that geometry with the current scene and remain sensitive to it during denoising. To address these challenges, we present MemCorr-DP, a diffusion policy that lifts frozen RoMa v2 matches into explicit 3D relations between the current scene and the reference trajectory. A counterfactual paired objective assigns opposite behaviors the same physical state and noisy action while retaining reference-specific denoising targets. Mixed-condition fine-tuning then adapts the policy from ground-truth geometry to measured correspondence errors. Our strongest evaluation places the Door in the outermost position bands beyond the training support and changes the query camera by $\pm15^\circ$. Under this combined shift, MemCorr-DP achieves 96.67% closed-loop success, compared with 88.00% for a visual Transformer with the same action architecture. Objective ablations and reference interventions show that behavior responds to the selected reference, while matched controls favor the complete relation set over future motion or centroid geometry alone. These results support explicit 3D reference relations as a robust conditioning interface when spatial and viewpoint changes are compounded in the evaluated task.
Abstract:Behavior-cloned visuomotor policies can remain accurate near their training distribution yet fail when object position and camera viewpoint change together. A successful reference trajectory contains the geometry needed to transfer the same interaction, but the policy must align that geometry with the current scene and remain sensitive to it during denoising. To address these challenges, we present MemCorr-DP, a diffusion policy that lifts frozen RoMa v2 matches into explicit 3D relations between the current scene and the reference trajectory. A counterfactual paired objective assigns opposite behaviors the same physical state and noisy action while retaining reference-specific denoising targets. Mixed-condition fine-tuning then adapts the policy from ground-truth geometry to measured correspondence errors. Our strongest evaluation places the Door in the outermost position bands beyond the training support and changes the query camera by $\pm15^\circ$. Under this combined shift, MemCorr-DP achieves 96.67% closed-loop success, compared with 88.00% for a visual Transformer with the same action architecture. Objective ablations and reference interventions show that behavior responds to the selected reference, while matched controls favor the complete relation set over future motion or centroid geometry alone. These results support explicit 3D reference relations as a robust conditioning interface when spatial and viewpoint changes are compounded in the evaluated task.
12 pages, 6 figures. Tan Su, Haoxiang Yang, and Ruxin Wang contributed equally. Corresponding author: Binghui Xie
Robotics (cs.RO); Artificial Intelligence (cs.AI)
arXiv:2609.06615 [cs.RO]
(or arXiv:2609.06615v1 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2609.06615 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 318 words
Related coverage
Deconfusing Corrigibility
AI Alignment Forum · Oct 17, 2026
Build a Physical AI model factory with NVIDIA Cosmos 3 on SageMaker HyperPod
AWS · Sep 4, 2026
E$^2$-OPSD: Taming Entropy Overshoot in On-Policy Self-Distillation
arXiv · Oct 6, 2026
Elite-Weighted Supervised Fine-tuning for Goal-Directed Molecular Optimization
arXiv · Sep 2, 2026