This repo provides code for the IEEE ICRA 2024 paper - Soft Fixtures: Practical Caging-Based Manipulation of Rigid and Deformable Objects

[PAPER] [WEBSITE] [VIDEOS]

Installation instructions:

Tested with:
Python 3.10
Ubuntu 22.04

0. Create a virtual environment (optional but recommended)

You can create a virtual environment using tools like virtualenv or conda.

1. Install OMPL with its Python bindings

Please first refer to the official installation instructions

For Ubuntu systems, you could simply do:

1.1 Download the OMPL installation script

1.2 Make it executable:

chmod u+x install-ompl-ubuntu.sh

1.3 Install the latest release of OMPL.app with Python bindings:

./install-ompl-ubuntu.sh --app

You can also choose to install from source.

2. Install other dependencies

pip install -r requirements.txt

Usage

1. Run escape energy analysis over a scenario

Here is a table of arguments you could use to customize your algorithm.

Argument	Short Option	Long Option	Default Value	Choices	Description
scenario	-c	--scenario	ShovelFish	ShovelFish, MaskEar, StarfishBowl, HookFishHole, HandbagGripper, BimanualRubic	Specify the scenario for the demo.
planner	-p	--planner	BITstar	BITstar, InformedRRTstar, RRTstar, RRT, AITstar	Specify the optimal planner to use.
object	-j	--object	Fish	Fish, MaskBand, Starfish, FishHole, Chain, Rubic	Specify the object to cage.
obstacle	-l	--obstacle	Shovel	Shovel, Ear, LeftHandAndBowl, Hook, 3fGripper	Specify the obstacle that cages the object.
runtime	-t	--runtime	30	N/A	Specify the runtime in seconds. Must be greater than 0.
visualization	-v	--visualization	True	False, True	Specify whether to visualize the PyBullet GUI.

A sample command is provided as follows,

python3 src/runScenario.py -c ShovelFish -p BITstar -j Fish -l Shovel -t 30 -v 1 

After running it, you will obtain results as following,

Please follow the correspondance in the arguments as below,

Scenario	Object	Obstacle
ShovelFish	Fish	Shovel
MaskEar	MaskBand	Ear
StarfishBowl	Starfish	LeftHandAndBowl
HookFishHole	FishHole	Hook
HandbagGripper	Chain	3fGripper
BimanualRubic	Rubic	3fGripper

2. Physical experiment

To reproduce our results on a Emika Franka Panda 7-axis robot arm, please

2.1 Refer to Franka FCI instructions and build libfranka

2.2 Copy the cpp script src/franka/generate_cartesian_pose_random_shaking.cpp to the path and add it to CMakeLists.txt

2.3 Rebuild the project

cd build
cmake -DCMAKE_BUILD_TYPE=Release -DBUILD_TESTS=OFF ..
cmake --build .

2.4 Run the script

~/libfranka/build$ ./examples/generate_cartesian_pose_motion <robot-ip>

Citation

If you find the code useful, please consider cite the following paper

@inproceedings{dong2024quasi,
  title={Quasi-static soft fixture analysis of rigid and deformable objects},
  author={Dong, Yifei and Pokorny, Florian T},
  booktitle={2024 IEEE International Conference on Robotics and Automation (ICRA)},
  pages={6513--6520},
  year={2024},
  organization={IEEE}
}

Acknowledgement

The code is partly built upon the repo - pybullet_ompl, which provides interface to use OMPL for motion planning inside PyBullet.

Contact or Support

Please contact Yifei Dong at [email protected] or Prof. Florian Pokorny at [email protected]

Contributors

The authors are with the division of Robotics, Perception and Learning, KTH Royal Institute of Technology, 100 44 Stockholm, Sweden. Funded by the European Commission under the Horizon Europe Framework Programme project SoftEnable, grant number 101070600.