Action Map Policy: Learning 3D Closed-loop Manipulation via Pixel Classification

• URL: http://arxiv.org/abs/2607.10706v1
• Date: Sun, 12 Jul 2026 10:58:06 GMT
• Title: Action Map Policy: Learning 3D Closed-loop Manipulation via Pixel Classification
• Abstract summary: Action Map Policy casts 3D closed-loop manipulation policy learning as a classification problem in image space.<n>We show that AMP outperforms strong baselines, achieving higher success rates, faster inference, and enhanced spatial reasoning.
• Score: 22.850651559843104
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: The action space poses a major challenge in robot learning, since it is often high-dimensional, can span long time horizons, and frequently admits multi-modal optimal solutions. A good choice of action representation and loss function can help to address these concerns, but there are often trade offs. We propose Action Map Policy (AMP), which casts 3D closed-loop manipulation policy learning as a classification problem in image space. While classification has been an effective formulation in generative language models, applying it to robot action learning is difficult because naively discretizing high-dimensional continuous actions explodes the token vocabulary. Our key idea is to project 3D actions onto the camera image planes and treat each pixel location as a discrete class, thus controlling dimensionality while retaining multi-modality. This method supports millimeter-level precision for high-dimensional actions without requiring a prohibitively large vocabulary, while preserving fine-grained pixel-wise visual signals. Furthermore, it can predict the entire action chunk in a single forward pass, avoiding complex noise scheduling and iterative denoising while achieving substantially faster inference than diffusion policies. Experiments on various manipulation tasks show that AMP outperforms strong baselines, achieving higher success rates, faster inference, and enhanced spatial reasoning.

Related papers

• Learning Action Manifold with Multi-view Latent Priors for Robotic Manipulation [67.16145181790522]
  This paper tackles spatial perception and manipulation challenges in Vision-Language-Action (VLA) models.<n>We propose a Geometry-Guided Gated Transformer (G3T) that aligns multi-view features under 3D geometric guidance.<n>To improve action learning efficiency, we introduce Action Manifold Learning (AML), which directly predicts actions on the valid action manifold.
  arXiv  Detail & Related papers  (2026-05-12T09:21:29Z)
• VolumeDP: Modeling Volumetric Representation for Manipulation Policy Learning [29.62906091681386]
  VolumeDP is a policy architecture that restores spatial alignment by explicitly reasoning in 3D.<n>It achieves a state-of-the-art average success rate of 88.8% on the LIBERO simulation benchmark.
  arXiv  Detail & Related papers  (2026-03-18T13:40:24Z)
• CLAMP: Contrastive Learning for 3D Multi-View Action-Conditioned Robotic Manipulation Pretraining [4.039082584778385]
  We introduce Contrastive Learning for 3D Multi-View Action-Conditioned Robotic Manipulation Pretraining (CLAMP)<n>From the merged point cloud computed from RGB-D images and camera extrinsics, we re-render multi-view four-channel image observations with depth and 3D coordinates.<n>The pre-trained encoders learn to associate the 3D geometric and positional information of objects with robot action patterns.
  arXiv  Detail & Related papers  (2026-01-31T23:32:54Z)
• DynaRend: Learning 3D Dynamics via Masked Future Rendering for Robotic Manipulation [52.136378691610524]
  We present DynaRend, a representation learning framework that learns 3D-aware and dynamics-informed triplane features.<n>By pretraining on multi-view RGB-D video data, DynaRend jointly captures spatial geometry, future dynamics, and task semantics in a unified triplane representation.<n>We evaluate DynaRend on two challenging benchmarks, RLBench and Colosseum, demonstrating substantial improvements in policy success rate, generalization to environmental perturbations, and real-world applicability across diverse manipulation tasks.
  arXiv  Detail & Related papers  (2025-10-28T10:17:11Z)
• Object-centric 3D Motion Field for Robot Learning from Human Videos [56.9436352861611]
  We propose to use object-centric 3D motion field to represent actions for robot learning from human videos.<n>We present a novel framework for extracting this representation from videos for zero-shot control.<n> Experiments show that our method reduces 3D motion estimation error by over 50% compared to the latest method.
  arXiv  Detail & Related papers  (2025-06-04T17:59:06Z)
• LLMI3D: MLLM-based 3D Perception from a Single 2D Image [77.13869413871028]
  multimodal large language models (MLLMs) excel in general capacity but underperform in 3D tasks.<n>In this paper, we propose solutions for weak 3D local spatial object perception, poor text-based geometric numerical output, and inability to handle camera focal variations.<n>We employ parameter-efficient fine-tuning for a pre-trained MLLM and develop LLMI3D, a powerful 3D perception MLLM.
  arXiv  Detail & Related papers  (2024-08-14T10:00:16Z)
• 3D Diffusion Policy: Generalizable Visuomotor Policy Learning via Simple 3D Representations [19.41216557646392]
  3D Diffusion Policy (DP3) is a novel visual imitation learning approach. In experiments, DP3 handles most tasks with just 10 demonstrations and surpasses baselines with a 24.2% relative improvement. In real robot experiments, DP3 rarely violates safety requirements, in contrast to baseline methods which frequently do.
  arXiv  Detail & Related papers  (2024-03-06T18:58:49Z)
• PolarNet: 3D Point Clouds for Language-Guided Robotic Manipulation [93.46306666726969]
  PolarNet is a 3D point cloud based policy for language-guided manipulation. It learns 3D point cloud representations and integrate them with language instructions for action prediction. It outperforms state-of-the-art 2D and 3D approaches in both single-task and multi-task learning.
  arXiv  Detail & Related papers  (2023-09-27T11:50:43Z)
• LocATe: End-to-end Localization of Actions in 3D with Transformers [91.28982770522329]
  LocATe is an end-to-end approach that jointly localizes and recognizes actions in a 3D sequence. Unlike transformer-based object-detection and classification models which consider image or patch features as input, LocATe's transformer model is capable of capturing long-term correlations between actions in a sequence. We introduce a new, challenging, and more realistic benchmark dataset, BABEL-TAL-20 (BT20), where the performance of state-of-the-art methods is significantly worse.
  arXiv  Detail & Related papers  (2022-03-21T03:35:32Z)
• Unsupervised Learning of Visual 3D Keypoints for Control [104.92063943162896]
  Learning sensorimotor control policies from high-dimensional images crucially relies on the quality of the underlying visual representations. We propose a framework to learn such a 3D geometric structure directly from images in an end-to-end unsupervised manner. These discovered 3D keypoints tend to meaningfully capture robot joints as well as object movements in a consistent manner across both time and 3D space.
  arXiv  Detail & Related papers  (2021-06-14T17:59:59Z)

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.