QDP: Learning to Sequentially Optimise Quasi-Static and Dynamic
Manipulation Primitives for Robotic Cloth Manipulation
Pre-defined manipulation primitives are widely used for cloth manipulation. However, cloth properties such as its stiffness or density can highly impact the performance of these primitives. Although existing solutions have tackled the parameterisation of pick and place locations, the effect of factors such as the velocity or trajectory of quasi-static and dynamic manipulation primitives has been neglected. Choosing appropriate values for these parameters is crucial to cope with the range of materials present in house-hold cloth objects.
To address this challenge, we introduce the Quasi-Dynamic Parameterisable (QDP) method, which optimises parameters such as the motion velocity in addition to the pick and place positions of quasi-static and dynamic manipulation primitives. In this work, we leverage the framework of Sequential Reinforcement Learning to decouple sequentially the parameters that compose the primitives. To evaluate the effectiveness of the method we focus on the task of cloth unfolding with a robotic arm in simulation and real-world experiments. Our results in simulation show that by deciding the optimal parameters for the primitives the performance can improve by 20% compared to sub-optimal ones. Real-world results demonstrate the advantage of modifying the velocity and height of manipulation primitives for cloths with different mass, stiffness, shape and size.
QDP proposes a sequence of sub-actions to find the optimal parameter values of quasi-static and dynamic manipulation primitives.
Short descriptive video
Content
We evaluate QDP on two simulation data sets and in the real world using cloths from the public household cloth object set.
We use three manipulation primitives:
Dynamic Quintic Polynomial: dynamic manipulation primitive. The primitive follows a semi-circle shaped trajectory from the pick to the place positions. The primitive is defined by its desired velocity at the middle point of the trajectory.
Pick-and-Place: quasi-static manipulation primitive. The primitive is defined by the height at which the cloth is lifted and the time to move from the pick to the place locations.
Drag: quasi-static manipulation primitive. The primitive can be considered as a pick-and-place primitive with zero height.
Real-World Experiments
Cotton Napkin (50 x 50 cm.)
Drag
Coverage: 35.58% -> 51.31%
P-n-P modifying the height
Coverage: 28.57% -> 59.89%
Dynamic Quintic Polynomial
Coverage: 32.08% -> 45.73%
Small Towel (30 x 50 cm.)
Videos at x5 speed
Drag
Coverage: 37.41% -> 40.31%
P-n-P modifying the height
Coverage: 33.67% ->84.56%
Dynamic Quintic Polynomial
Coverage: 36.61% -> 50.20%
Chequered Rag (50 x 70 cm.)
Coming Soon
Simulation Experiments
Normal Size Cloths
Manipulation Primitive: Drag
Final cloth coverage: 100%
Final cloth coverage:100%
Manipulation Primitive: Pick-and-Place
Modifying the height parameter:
Final cloth coverage: 79.10%
Final cloth coverage: 56.75%
Modifying the time parameter:
Final cloth coverage: 77.59%
Final cloth coverage: 73.04%
Manipulation Primitive: Dynamic Quintic Polynomial
Final cloth coverage: 96.94%
Final cloth coverage: 89.49%
Large Size Cloths
Manipulation Primitive: Drag
Final cloth coverage: 81.37%
Final cloth coverage:89.16%
Manipulation Primitive: Pick-and-Place
Modifying the height parameter:
Final cloth coverage: 84.27%
Final cloth coverage: 82.31%
Modifying the time parameter:
Final cloth coverage: 82.08%
Final cloth coverage: 49.71%
Manipulation Primitive: Dynamic Quintic Polynomial
Final cloth coverage: 93.50%
Final cloth coverage: 78.03%
Team
Munich Institute of Robotics and Machine Intelligence
Technische Universität München
Citation
To cite this work, please use the following BibTex entry:
@inproceedings{blancomulero2023qdp,
author={Blanco-Mulero, David and Alcan, Gokhan and Abu-Dakka, Fares J. and Kyrki, Ville},
booktitle={2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)},
title={QDP: Learning to Sequentially Optimise Quasi-Static and Dynamic Manipulation Primitives for Robotic Cloth Manipulation},
year={2023},
volume={},
number={},
pages={984-991},
doi={10.1109/IROS55552.2023.10342002}
}