A project by IPPLAB · Deformable Manipulation

Robots learning to manipulate
the soft world

A curated map of deformable simulation and manipulation for embodied AI: perception, control, policy learning, sim-to-real transfer, and asset generation for cloth, rope, and soft bodies.

Schematic diagram of robot gripper lifting flexible fabric
Deformable Manipulation research
08 Research areas
Open source · Continuously curated

DEFORM · SIMULATE · LEARN

01 / OVERVIEW

From the physical world
to intelligent policies

Flexible objects have no fixed shape, but exist in every real scene that robots need to face.

Explore papers, tools, datasets, and benchmarks across perception, simulation, manipulation, and transfer.

01ModelingModel
02SimulateSimulate
03LearnLearn
04TransferTransfer

02 / RESEARCH MAP

Eight directions,one map

Explore deformable simulation and manipulation, from numerical solvers to real-world deployment.

02◇

MANIPULATION

Deformable manipulation tasks

Task, data and policy learning for folding, hanging, grabbing and moving operations.

View resources ↗
03◇

MANIPULATION METHODS

Deformable manipulation methods

From grasp point perception, visual and tactile feedback to general garment manipulation and strategy learning.

View resources ↗
04⇄

SIM-TO-REAL

Sim-to-real transfer

Narrow the gap between dynamics and visual domains, allowing simulation training to truly reach the real world.

View resources ↗
05▦

BENCHMARKS

Benchmarks and evaluation

Use standard environments and unified indicators to reliably compare the learning efficiency and generalization ability of algorithms.

View resources ↗
06✦

3D GENERATION

Physical asset generation

Reconstruct simulatable, physically-property 3D flexible assets from images or videos.

View resources ↗
07∿

FOUNDATIONS

Simulation foundations

PBD, finite elements and differentiable physics: understanding the computational methods behind the motion of deformable bodies.

View resources ↗
08▤

SURVEYS

Surveys and reading guides

Establish one map of the field of deformable objects and cloth manipulation, and sort out the methodological context and open issues.

View resources ↗

03 / HIGHLIGHTS

Start here

View full list →

04 / OPEN COLLABORATION

Help map the
deformable world

Found a paper, tool, or dataset worth including? Feel free to submit Pull Requests to keep this list growing.