Disentangled Point Diffusion for Precise Object Placement
2026-04-13 • Robotics
Robotics
AI summaryⓘ
The authors propose TAX-DPD, a new method to help robots place objects precisely by predicting where to put them in a scene. Their approach breaks down the problem into global placement prediction and local object configuration using a special point cloud diffusion technique. This method works better than previous ones at handling different object shapes and positions, and it was tested successfully on precise industrial tasks and cloth hanging in simulations. The authors show that their approach improves accuracy and flexibility compared to older diffusion methods.
robotic manipulationlearning from demonstrationpoint diffusionGaussian Mixture Modelobject placementpose predictionpoint cloudSE(3) diffusionhierarchical modelsrigid and non-rigid objects
Authors
Lyuxing He, Eric Cai, Shobhit Aggarwal, Jianjun Wang, David Held
Abstract
Recent advances in robotic manipulation have highlighted the effectiveness of learning from demonstration. However, while end-to-end policies excel in expressivity and flexibility, they struggle both in generalizing to novel object geometries and in attaining a high degree of precision. An alternative, object-centric approach frames the task as predicting the placement pose of the target object, providing a modular decomposition of the problem. Building on this goal-prediction paradigm, we propose TAX-DPD, a hierarchical, disentangled point diffusion framework that achieves state-of-the-art performance in placement precision, multi-modal coverage, and generalization to variations in object geometries and scene configurations. We model global scene-level placements through a novel feed-forward Dense Gaussian Mixture Model (GMM) that yields a spatially dense prior over global placements; we then model the local object-level configuration through a novel disentangled point cloud diffusion module that separately diffuses the object geometry and the placement frame, enabling precise local geometric reasoning. Interestingly, we demonstrate that our point cloud diffusion achieves substantially higher accuracy than a prior approach based on SE(3)-diffusion, even in the context of rigid object placement. We validate our approach across a suite of challenging tasks in simulation and in the real-world on high-precision industrial insertion tasks. Furthermore, we present results on a cloth-hanging task in simulation, indicating that our framework can further relax assumptions on object rigidity.