# Emerging Extrinsic Dexterity in Cluttered Scenes via Dynamics-aware Policy Learning

arXiv: https://arxiv.org/abs/2603.09882 · HTML: https://mlsanity.com/paper/2603.09882/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 37 of its 45 references that are in the map.
- First posted on arXiv 10 Mar 2026.

## Builds on (most cited first, 30 of 37)

- [SAM 2: Segment Anything in Images and Videos](https://mlsanity.com/paper/2408.00714/) (2024)
- [Objaverse: A Universe of Annotated 3D Objects](https://mlsanity.com/paper/2212.08051/) (2022)
- [OpenAI GPT-5 System Card](https://mlsanity.com/paper/2601.03267/) (2025)
- [XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model](https://mlsanity.com/paper/2207.07115/) (2022)
- [FoundationPose: Unified 6D Pose Estimation and Tracking of Novel Objects](https://mlsanity.com/paper/2312.08344/) (2023)
- [GELLO: A General, Low-Cost, and Intuitive Teleoperation Framework for Robot Manipulators](https://mlsanity.com/paper/2309.13037/) (2023)
- [Where2Act: From Pixels to Actions for Articulated 3D Objects](https://mlsanity.com/paper/2101.02692/) (2021)
- [GraspVLA: a Grasping Foundation Model Pre-trained on Billion-scale Synthetic Action Data](https://mlsanity.com/paper/2505.03233/) (2025)
- [Sonata: Self-Supervised Learning of Reliable Point Representations](https://mlsanity.com/paper/2503.16429/) (2025)
- [ContactGrasp: Functional Multi-finger Grasp Synthesis from Contact](https://mlsanity.com/paper/1904.03754/) (2019)
- [PointNetGPD: Detecting Grasp Configurations from Point Sets](https://mlsanity.com/paper/1809.06267/) (2018)
- [Approximate Convex Decomposition for 3D Meshes with Collision-Aware Concavity and Tree Search](https://mlsanity.com/paper/2205.02961/) (2022)
- [DexArt: Benchmarking Generalizable Dexterous Manipulation with Articulated Objects](https://mlsanity.com/paper/2305.05706/) (2023)
- [SoFar: Language-Grounded Orientation Bridges Spatial Reasoning and Object Manipulation](https://mlsanity.com/paper/2502.13143/) (2025)
- [Concerto: Joint 2D-3D Self-Supervised Learning Emerges Spatial Representations](https://mlsanity.com/paper/2510.23607/) (2025)
- [FACTR: Force-Attending Curriculum Training for Contact-Rich Policy Learning](https://mlsanity.com/paper/2502.17432/) (2025)
- [GraspGen: A Diffusion-based Framework for 6-DOF Grasping with On-Generator Training](https://mlsanity.com/paper/2507.13097/) (2025)
- [Point2Vec for Self-Supervised Representation Learning on Point Clouds](https://mlsanity.com/paper/2303.16570/) (2023)
- [GENMANIP: LLM-driven Simulation for Generalizable Instruction-Following Manipulation](https://mlsanity.com/paper/2506.10966/) (2025)
- [Contact Mode Guided Motion Planning for Quasidynamic Dexterous Manipulation in 3D](https://mlsanity.com/paper/2105.14431/) (2021)
- [Pre- and post-contact policy decomposition for non-prehensile manipulation with zero-shot sim-to-real transfer](https://mlsanity.com/paper/2309.02754/) (2023)
- [Contact Mode Guided Sampling-Based Planning for Quasistatic Dexterous Manipulation in 2D](https://mlsanity.com/paper/2011.01454/) (2020)
- [Enhancing Dexterity in Robotic Manipulation via Hierarchical Contact Exploration](https://mlsanity.com/paper/2307.00383/) (2023)
- [In-Hand Manipulation via Motion Cones](https://mlsanity.com/paper/1810.00219/) (2018)
- [One-Shot Transfer of Long-Horizon Extrinsic Manipulation Through Contact Retargeting](https://mlsanity.com/paper/2404.07468/) (2024)
- [Sampling-based Planning of In-Hand Manipulation with External Pushes](https://mlsanity.com/paper/1707.00318/) (2017)
- [CORN: Contact-based Object Representation for Nonprehensile Manipulation of General Unseen Objects](https://mlsanity.com/paper/2403.10760/) (2024)
- [Pre-trained Visual Dynamics Representations for Efficient Policy Learning](https://mlsanity.com/paper/2411.03169/) (2024)
- [4D Visual Pre-training for Robot Learning](https://mlsanity.com/paper/2508.17230/) (2025)
- [Tactile-Driven Non-Prehensile Object Manipulation via Extrinsic Contact Mode Control](https://mlsanity.com/paper/2405.18214/) (2024)

## Built on by (most cited first, 2 of 2)

- [ProxiDex: Learning Dynamics-Guided Proximity Policy for Dexterous Manipulation](https://mlsanity.com/paper/2609.16586/) (2026)
- [DualWAM: Dual-System World Action Models for Asynchronous Global Planning and Local Refinement](https://mlsanity.com/paper/2609.24868/) (2026)

## Abstract (arXiv)

Extrinsic dexterity leverages environmental contact to overcome the limitations of prehensile manipulation. However, achieving such dexterity in cluttered scenes remains challenging and underexplored, as it requires selectively exploiting contact among multiple interacting objects with inherently coupled dynamics. Existing approaches lack explicit modeling of such complex dynamics and therefore fall short in non-prehensile manipulation in cluttered environments, which in turn limits their practical applicability in real-world environments. In this paper, we introduce a Dynamics-Aware Policy Learning (DAPL) framework that can facilitate policy learning with a learned representation of contact-induced object dynamics in cluttered environments. This representation is learned through explicit world modeling and used to condition reinforcement learning, enabling extrinsic dexterity to emerge without hand-crafted contact heuristics or complex reward shaping. We evaluate our approach in both simulation and the real world. Our method outperforms prehensile manipulation, human teleoperation, and prior representation-based policies by over 25% in success rate on unseen simulated cluttered scenes with varying densities. The real-world success rate reaches around 50% across 10 cluttered scenes, while a practical grocery deployment further demonstrates robust sim-to-real transfer and applicability.