# Universal Visual Decomposer: Long-Horizon Manipulation Made Easy

arXiv: https://arxiv.org/abs/2310.08581 · HTML: https://mlsanity.com/paper/2310.08581/

## Key facts

- Cited by 25 papers in the map (data through 14 Oct 2026).
- Builds on 55 of its 68 references that are in the map.
- First posted on arXiv 12 Oct 2023.

## Builds on (most cited first, 30 of 55)

- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [Llama 2: Open Foundation and Fine-Tuned Chat Models](https://mlsanity.com/paper/2307.09288/) (2023)
- [DINOv2: Learning Robust Visual Features without Supervision](https://mlsanity.com/paper/2304.07193/) (2023)
- [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://mlsanity.com/paper/2104.09864/) (2021)
- [Supervised Contrastive Learning](https://mlsanity.com/paper/2004.11362/) (2020)
- [Ego4D: Around the World in 3,000 Hours of Egocentric Video](https://mlsanity.com/paper/2110.07058/) (2021)
- [Decision Transformer: Reinforcement Learning via Sequence Modeling](https://mlsanity.com/paper/2106.01345/) (2021)
- [Root Mean Square Layer Normalization](https://mlsanity.com/paper/1910.07467/) (2019)
- [Scaling Egocentric Vision: The EPIC-KITCHENS Dataset](https://mlsanity.com/paper/1804.02748/) (2018)
- [R3M: A Universal Visual Representation for Robot Manipulation](https://mlsanity.com/paper/2203.12601/) (2022)
- [Perceiver-Actor: A Multi-Task Transformer for Robotic Manipulation](https://mlsanity.com/paper/2209.05451/) (2022)
- [Time-Contrastive Networks: Self-Supervised Learning from Video](https://mlsanity.com/paper/1704.06888/) (2017)
- [Data-Efficient Hierarchical Reinforcement Learning](https://mlsanity.com/paper/1805.08296/) (2018)
- [The Option-Critic Architecture](https://mlsanity.com/paper/1609.05140/) (2016)
- [VIP: Towards Universal Visual Reward and Representation via Value-Implicit Pre-Training](https://mlsanity.com/paper/2210.00030/) (2022)
- [Relay Policy Learning: Solving Long-Horizon Tasks via Imitation and Reinforcement Learning](https://mlsanity.com/paper/1910.11956/) (2019)
- [DD-PPO: Learning Near-Perfect PointGoal Navigators from 2.5 Billion Frames](https://mlsanity.com/paper/1911.00357/) (2019)
- [Learning Latent Plans from Play](https://mlsanity.com/paper/1903.01973/) (2019)
- [Behavior Transformers: Cloning k modes with one stone](https://mlsanity.com/paper/2206.11251/) (2022)
- [Masked Visual Pre-training for Motor Control](https://mlsanity.com/paper/2203.06173/) (2022)
- [Where are we in the search for an Artificial Visual Cortex for Embodied Intelligence?](https://mlsanity.com/paper/2303.18240/) (2023)
- [Search on the Replay Buffer: Bridging Planning and Reinforcement Learning](https://mlsanity.com/paper/1906.05253/) (2019)
- [RoboTHOR: An Open Simulation-to-Real Embodied AI Platform](https://mlsanity.com/paper/2004.06799/) (2020)
- [Simple but Effective: CLIP Embeddings for Embodied AI](https://mlsanity.com/paper/2111.09888/) (2021)
- [LIV: Language-Image Representations and Rewards for Robotic Control](https://mlsanity.com/paper/2306.00958/) (2023)
- [Goal-conditioned Imitation Learning](https://mlsanity.com/paper/1906.05838/) (2019)
- [The Unsurprising Effectiveness of Pre-Trained Vision Models for Control](https://mlsanity.com/paper/2203.03580/) (2022)
- [Planning with Goal-Conditioned Policies](https://mlsanity.com/paper/1911.08453/) (2019)

## Built on by (most cited first, 25 of 25)

- [A Single Goal is All You Need: Skills and Exploration Emerge from Contrastive RL without Rewards, Demonstrations, or Subgoals](https://mlsanity.com/paper/2408.05804/) (2024)
- [Mamba Policy: Towards Efficient 3D Diffusion Policy with Hybrid Selective State Models](https://mlsanity.com/paper/2409.07163/) (2024)
- [Learning Compositional Behaviors from Demonstration and Language](https://mlsanity.com/paper/2505.21981/) (2025)
- [Scaling Robot Policy Learning via Zero-Shot Labeling with Foundation Models](https://mlsanity.com/paper/2410.17772/) (2024)
- [Temporal Representation Alignment: Successor Features Enable Emergent Compositionality in Robot Instruction Following](https://mlsanity.com/paper/2502.05454/) (2025)
- [EXTRACT: Efficient Policy Learning by Extracting Transferable Robot Skills from Offline Data](https://mlsanity.com/paper/2406.17768/) (2024)
- [Learning from 10 Demos: Generalisable and Sample-Efficient Policy Learning with Oriented Affordance Frames](https://mlsanity.com/paper/2410.12124/) (2024)
- [Subtask-Aware Visual Reward Learning from Segmented Demonstrations](https://mlsanity.com/paper/2502.20630/) (2025)
- [KOI: Accelerating Online Imitation Learning via Hybrid Key-state Guidance](https://mlsanity.com/paper/2408.02912/) (2024)
- [Elastic Motion Policy: An Adaptive Dynamical System for Robust and Efficient One-Shot Imitation Learning](https://mlsanity.com/paper/2503.08029/) (2025)
- [Provable Ordering and Continuity in Vision-Language Pretraining for Generalizable Embodied Agents](https://mlsanity.com/paper/2502.01218/) (2025)
- [LodeStar: Long-horizon Dexterity via Synthetic Data Augmentation from Human Demonstrations](https://mlsanity.com/paper/2508.17547/) (2025)
- [Gaze-Guided Task Decomposition for Imitation Learning in Robotic Manipulation](https://mlsanity.com/paper/2501.15071/) (2025)
- [RDD: Retrieval-Based Demonstration Decomposer for Planner Alignment in Long-Horizon Tasks](https://mlsanity.com/paper/2510.14968/) (2025)
- [RARM: Confidence-Gated Progress Reward Modeling for RL in Manipulation](https://mlsanity.com/paper/2606.22027/) (2026)
- [GHOST: Hierarchical Sub-Goal Policies for Generalizing Robot Manipulation](https://mlsanity.com/paper/2606.10025/) (2026)
- [RAPID: Robot Agentic Programming from Demonstrations](https://mlsanity.com/paper/2609.30249/) (2026)
- [Flat'n'Fold: A Diverse Multi-Modal Dataset for Garment Perception and Manipulation](https://mlsanity.com/paper/2409.18297/) (2024)
- [Single-Shot Learning of Stable Dynamical Systems for Long-Horizon Manipulation Tasks](https://mlsanity.com/paper/2410.01033/) (2024)
- [Robust Offline Imitation Learning Through State-level Trajectory Stitching](https://mlsanity.com/paper/2503.22524/) (2025)
- [SAGE: Scene Graph-Aware Guidance and Execution for Long-Horizon Manipulation Tasks](https://mlsanity.com/paper/2509.21928/) (2025)
- [PrimitiveVLA: Learning Reusable Motion Primitives for Efficient and Generalizable Robotic Manipulation](https://mlsanity.com/paper/2605.28634/) (2026)
- [GORDON: Graph-based Object-centric Rewards for Decomposition of Long-Horizon Manipulation](https://mlsanity.com/paper/2608.03753/) (2026)
- [CASD: Chunk-Aligned Semantic Distillation for Multi-StageRobot Manipulation](https://mlsanity.com/paper/2609.08638/) (2026)
- [SeeQ: Training Generalist Value Functions for Long-Horizon Robotic Manipulation](https://mlsanity.com/paper/2609.22085/) (2026)

## Abstract (arXiv)

Real-world robotic tasks stretch over extended horizons and encompass multiple stages. Learning long-horizon manipulation tasks, however, is a long-standing challenge, and demands decomposing the overarching task into several manageable subtasks to facilitate policy learning and generalization to unseen tasks. Prior task decomposition methods require task-specific knowledge, are computationally intensive, and cannot readily be applied to new tasks. To address these shortcomings, we propose Universal Visual Decomposer (UVD), an off-the-shelf task decomposition method for visual long horizon manipulation using pre-trained visual representations designed for robotic control. At a high level, UVD discovers subgoals by detecting phase shifts in the embedding space of the pre-trained representation. Operating purely on visual demonstrations without auxiliary information, UVD can effectively extract visual subgoals embedded in the videos, while incurring zero additional training cost on top of standard visuomotor policy training. Goal-conditioned policies learned with UVD-discovered subgoals exhibit significantly improved compositional generalization at test time to unseen tasks. Furthermore, UVD-discovered subgoals can be used to construct goal-based reward shaping that jump-starts temporally extended exploration for reinforcement learning. We extensively evaluate UVD on both simulation and real-world tasks, and in all cases, UVD substantially outperforms baselines across imitation and reinforcement learning settings on in-domain and out-of-domain task sequences alike, validating the clear advantage of automated visual task decomposition within the simple, compact UVD framework.