![Figure 2.12: PR2 robot performing a folding sequence using the method presented in [7]. On the left, the initial crumpled state is shown. Through a sequence of manipulations the robot is able to unfold the pants to a known state (shown on the right), and the robot can then proceed to fold them. The figure was originally published as](/figures/figure-2-12-pr2-robot-performing-a-folding-sequence-using-38p0gm4z.webp)
Q2. What future works have the authors mentioned in the paper "A n d m a n i p u l at i o n david estévez fernández a dissertation submitted by in partial fulfillment of the requirements for the degree of doctor of philosophy in electrical engineering, electronics and automation universidad carlos iii de madrid advisors: carlos balaguer bernaldo de quirós juan carlos gonzález víctores tutor: juan carlos gonzález víctores" ?
141 8. 2 Future Lines of Work................. 142 8. 2. 1 Hanging.................... 142 8. 2. 2 Unfolding................... 143 8. 2. 3 Ironing..................... 144 contents xix 8. 2. 4 Folding..................... 145 bibliography 147 motion.................... 5 Figure 1. 2 Garment deformability as a challenge for perception.................. 6 Figure 1. 3 Garment deformability as a challenge for manipulation................. 7 Figure 1. 4 Proposed laundry pipeline......... 10 Figure 2. 1 Isolation task................. In addition, deformability creates the possibility of self occlusions, as not only external objects or other clothing articles can occlude the garment, but parts of same garment can prevent the camera to obtain a view of certain parts of itself. The proposed methods and techniques developed will be subjected to experimental validation and critical review of the results obtained to evaluate the degree of success. To achieve this objective, a synthetic dataset will be created to study how a robotic system can predict the hangability of a given garment when dropped over a hanger.
Q3. What is the method used to remove marked creases?
Once located, marked creases are removed by static or dynamic ironing using a combination of position control in the robot and the use of a foam under the cloth as a source of passive compliance.
Q4. What are the common operations with garments?
Typical operations with garments involve separating them from other clothes or moving specific parts of the garment, such as overlapping folds.
Q5. What is the effect of the location of the garment on the current state of the garment?
As garments are deformable objects, the location of the garment (over a flat surface, hanging on a rope, etc) greatly influences the current garment state.
Q6. What is the method used to detect the upper layer of a garment?
The upper layer is detected from a depth image, through a depth first algorithm and a simple perceptron applied on a simplified action space based on the edges detected on the garment input image.
Q7. What is the main reason garments are deformable?
In addition, deformability creates the possibility of self occlusions, as not only external objects or other clothing articles can occlude the garment, but parts of same garment can prevent the camera to obtain a view of certain parts of itself.
Q8. What is the method used to perceive a garment?
To perceive the clothing article, it combines 3D information from a 360º stereo scan with 2D information from several views, using fiducial markers on the clothing article to determine the garment state.
Q9. Why is it unfeasible to compile a dataset with tens?
Due to the large amount of time that has to be devoted to each individual example, compiling a dataset with tens of thousands of training examples is unattainable unless several robots are setup to work in parallel, which is unfeasible for the budget of most research groups.
Q10. How can garment state estimation be performed?
The method2.4 garment state estimation 27is able to generate the corresponding mesh automatically and handle in-plane rotation by re-initializing the mesh after data has been lost in the image sequence.
Q11. What is the way to manipulate garments?
successful garment manipulation ideally requires either robotic hands equipped with sensors and fine manipulation skills, or specialized tools specifically designed to handle them correctly.
Q12. How do they estimate the shape of a deformable object?
Han et al. [24] introduce a method to estimate the shape of a deformable object from a RGB-D image sequence using Signed28 backgroundDistance Functions (SDFs), to use it in the feedback loop of a manipulation controller.
Q13. How were the paths learned from user demonstrations?
Ironing paths were learned from user demonstrations by kinesthetic teaching (Figure 2.6), and force profiles were extracted from demonstrations via a haptic device.
Q14. What is the main advantage of a robotic system over a traditional one?
Although the throughput of such a system would probably be lower, the main advantage of installing a robotic system versus a traditional automated system is the increased adaptability to different garments and tasks.