mlsanityData to 14 Oct 2026

Dynamics Learning with Cascaded Variational Inference for Multi-Step Manipulation

arXiv 1910.13395 ↗First posted 29 Oct 2019cs.RO cs.CV cs.LGcited by 24

Not on the map. Why?

Abstract (arXiv)

The fundamental challenge of planning for multi-step manipulation is to find effective and plausible action sequences that lead to the task goal. We present Cascaded Variational Inference (CAVIN) Planner, a model-based method that hierarchically generates plans by sampling from latent spaces. To facilitate planning over long time horizons, our method learns latent representations that decouple the prediction of high-level effects from the generation of low-level motions through cascaded variational inference. This enables us to model dynamics at two different levels of temporal resolutions for hierarchical planning. We evaluate our approach in three multi-step robotic manipulation tasks in cluttered tabletop environments given high-dimensional observations. Empirical results demonstrate that the proposed method outperforms state-of-the-art model-based methods by strategically interacting with multiple objects.

Key facts. Cited by 24 papers in the map (data through 14 Oct 2026). First posted on arXiv 29 Oct 2019.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2020: 1; 2021: 5; 2022: 6; 2023: 5; 2024: 1; 2025: 2; 2026: 4.

24 papers.

Sort:

By area:

  1. 2021ManipulaTHOR: A Framework for Visual Object Manipulationcited by 102
  2. 2022Latent Plans for Task-Agnostic Offline Reinforcement Learningcited by 81
  3. 2020Deep Imitation Learning for Bimanual Robotic Manipulationcited by 50
  4. 2026Hierarchical Planning with Latent World Modelscited by 43
  5. 2021LASER: Learning a Latent Action Space for Efficient Reinforcement Learningcited by 32
  6. 2022Planning to Practice: Efficient Online Fine-Tuning by Composing Goals in Latent Spacecited by 31
  7. 2022Generalization with Lossy Affordances: Leveraging Broad Offline Data for Learning Visuomotor Taskscited by 25
  8. 2023DexDeform: Dexterous Deformable Object Manipulation with Human Demonstrations and Differentiable Physicscited by 23
  9. 2021Subgoal Search For Complex Reasoning Taskscited by 21
  10. 2023Reparameterized Policy Learning for Multimodal Trajectory Optimizationcited by 20
  11. 2025Temporal Representation Alignment: Successor Features Enable Emergent Compositionality in Robot Instruction Followingcited by 14
  12. 2021Closing the Planning-Learning Loop with Application to Autonomous Drivingcited by 12
12 more
  1. 2022Fast and Precise: Adjusting Planning Horizon with Adaptive Subgoal Searchcited by 12
  2. 2023Action Dynamics Task Graphs for Learning Plannable Representations of Procedural Taskscited by 8
  3. 2024ActionDiffusion: An Action-aware Diffusion Model for Procedure Planning in Instructional Videoscited by 8
  4. 2023Hybrid Search for Efficient Planning with Completeness Guaranteescited by 6
  5. 2023DiffVL: Scaling Up Soft Body Manipulation using Vision-Language Driven Differentiable Physicscited by 6
  6. 2021Discovering Generalizable Skills via Automated Generation of Diverse Taskscited by 5
  7. 2022Active Task Randomization: Learning Robust Skills via Unsupervised Generation of Diverse and Feasible Taskscited by 3
  8. 2022STEADY: Simultaneous State Estimation and Dynamics Learning from Indirect Observationscited by 2
  9. 2026Unifying Object-Centric World Models and Diffusion Policy: A Hierarchical Framework for Multi-Stage Robotic Taskscited by 2
  10. 2025Correspondence-Oriented Imitation Learning: Flexible Visuomotor Control with 3D Conditioningcited by 1
  11. 2026Flow Policies as Actions of Skill-Level World Models: Learned and Symbolic Abstractions for Long-Horizon Planningcited by 0
  12. 2026H-JEPA: End-to-End Learning of Hierarchical World Models for Visual Planningcited by 0