# Learning 3D Affordances for Blade Insertion in Cluttered Stowing

arXiv: https://arxiv.org/abs/2607.02549 · HTML: https://mlsanity.com/paper/2607.02549/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 22 of its 24 references that are in the map.
- First posted on arXiv 25 Jun 2026.

## Builds on (most cited first, 22 of 22)

- [Masked Autoencoders Are Scalable Vision Learners](https://mlsanity.com/paper/2111.06377/) (2021)
- [Emerging Properties in Self-Supervised Vision Transformers](https://mlsanity.com/paper/2104.14294/) (2021)
- [V-Net: Fully Convolutional Neural Networks for Volumetric Medical Image Segmentation](https://mlsanity.com/paper/1606.04797/) (2016)
- [Occupancy Networks: Learning 3D Reconstruction in Function Space](https://mlsanity.com/paper/1812.03828/) (2018)
- [MonoScene: Monocular 3D Semantic Scene Completion](https://mlsanity.com/paper/2112.00726/) (2021)
- [Transporter Networks: Rearranging the Visual World for Robotic Manipulation](https://mlsanity.com/paper/2010.14406/) (2020)
- [Mask DINO: Towards A Unified Transformer-based Framework for Object Detection and Segmentation](https://mlsanity.com/paper/2206.02777/) (2022)
- [3D Diffuser Actor: Policy Diffusion with 3D Scene Representations](https://mlsanity.com/paper/2402.10885/) (2024)
- [Affordances from Human Videos as a Versatile Representation for Robotics](https://mlsanity.com/paper/2304.08488/) (2023)
- [Neural Descriptor Fields: SE(3)-Equivariant Object Representations for Manipulation](https://mlsanity.com/paper/2112.05124/) (2021)
- [Where2Act: From Pixels to Actions for Articulated 3D Objects](https://mlsanity.com/paper/2101.02692/) (2021)
- [VAT-Mart: Learning Visual Action Trajectory Proposals for Manipulating 3D ARTiculated Objects](https://mlsanity.com/paper/2106.14440/) (2021)
- [D3Fields: Dynamic 3D Descriptor Fields for Zero-Shot Generalizable Rearrangement](https://mlsanity.com/paper/2309.16118/) (2023)
- [Learning Environment-Aware Affordance for 3D Articulated Object Manipulation under Occlusions](https://mlsanity.com/paper/2309.07510/) (2023)
- [UAD: Unsupervised Affordance Distillation for Generalization in Robotic Manipulation](https://mlsanity.com/paper/2506.09284/) (2025)
- [3D FlowMatch Actor: Unified 3D Policy for Single- and Dual-Arm Manipulation](https://mlsanity.com/paper/2508.11002/) (2025)
- [GIFT: Generalizable Interaction-aware Functional Tool Affordances without Labels](https://mlsanity.com/paper/2106.14973/) (2021)
- [CabiNet: Scaling Neural Collision Detection for Object Rearrangement with Procedural Scene Generation](https://mlsanity.com/paper/2304.09302/) (2023)
- [Adapt3R: Adaptive 3D Scene Representation for Domain Transfer in Imitation Learning](https://mlsanity.com/paper/2503.04877/) (2025)
- [3D-MVP: 3D Multiview Pretraining for Robotic Manipulation](https://mlsanity.com/paper/2406.18158/) (2024)
- [Stow: Robotic Packing of Items into Fabric Pods](https://mlsanity.com/paper/2505.04572/) (2025)
- [SAGA: Open-World Mobile Manipulation via Structured Affordance Grounding](https://mlsanity.com/paper/2512.12842/) (2025)

## Abstract (arXiv)

Many manipulation tasks require reasoning about free-space affordances: discovering volumes where an extended rigid tool can safely navigate, complementary to surface contact affordances for grasping. Robotic stowing is a canonical instance, where a blade must sweep items aside inside cluttered fabric bins to create insertion space. Production stow systems generate millions of such episodes, but standard approaches with unimodal data infer affordances as SE(3) pose distributions, a geometric question asked in the wrong domain. VulcanVoxel keeps inference spatial: a masked autoencoder over 3D occupancy fields reconstructs blade occupancy conditioned on scene geometry, computing feasibility locally at each voxel and recovering multi-modal predictions from unimodal data. Blade affordances are spatial objects, subsets of 3D space defined by geometric feasibility. Pose parameters carry no structure for reasoning whether unobserved placements are feasible, and standard generative objectives including flow matching faithfully learn the unimodal distribution produced by execution policies and cannot recover geometric alternatives. Trained on 10,000 real warehouse stow episodes without human annotation, VulcanVoxel achieves top-5 coverage of 0.89 versus 0.71 for the best pose-based baseline, with a distilled student providing RGB-to-voxel inference in 30 ms. vs. 1.4 s. for voxel-to-voxel. We have released a dataset of real blade insertion cycles with RGB-D observations and pose trajectories at https://www.armbench.com/blade_insertion. html.