A CloPeMa team will run an ICRA Workshop on Advances in Robot Manipulation of Clothes and Flexible Objects on 1st June 2014 in Hong Kong. The problem of manipulating and handling deformable and flexible objects is a great challenge in the development of robots capable of performing actions usually carried out by humans. It holds a great variety of technical challenges in new and advanced modelling, sensing, actuation planning and control technologies.
This ICRA workshop is aimed at the aspects related to the manipulation of flexible objects in the areas of: vision, planning, grasping, manipulation and interaction control, haptics, human-robot interaction, cognitive robotics, AI and others. It will, for instance, address technological aspects such as gripper design for advanced manipulation of flexible objects and visual and touch based coordinated robot control for execution of complex tasks. Furthermore it will give detailed insights into the cognitive, sensing and modelling issues in manipulation of flexible objects and is a chance to hear about and discuss the latest advancements.
The ICRA workshop is a great opportunity to meet and talk to the experts, such as, Sotiris Malassiotis, Vaclav Hlavak, Giorgio Cannata, Andreas Doumanoglou and other CloPeMa members. Register here.
Also the paper Autonomous Active Recognition and Unfolding of Clothes using Random Decision Forests and Probabilistic Planning will be presented in the main conference. Authors: Andreas Doumanoglou, Andreas Kargakos, Tae-Kyun Kim, Sotiris Malassiotis
We had the great opportunity to present our research to a variety of industry partners on the 5th of November 2013, at the SICSA Demofest. SICSA stands for Scottish Informatics and Computer Science Alliance, who created the Demofest as an event to introduce industry to ongoing academic research. It is a chance to find industry partners who are interested in investing and collaborating in cutting edge research around the country.
The Glasgow Stereo Image Database of Garments is a collection of 80 stereo-pair colour images of garments with corresponding horizontal and vertical disparity maps and mask annotations. This database is based on 16 different off-the-shelve garments. Each garment has been imaged in five different pose configurations using the CloPeMa active binocular robot head. The aim of this database is to serve as a benchmark tool for algorithms for recognition, segmentation and various range image properties of non-rigid objects.
The compiler works as a cross compiler on the host machine. The compiler flag you must use is -cpuMIC to generate code for it. You need to have the Intel MPSS package installed and you need to set up a pair of symbolic links on your path to the MIC assembler and MIC gcc compilers. The Pascal compiler expects a MIC version of the GNU assembler to be on its path with the name mic-as and a mic version of the gcc compiler with the name mic-gcc.
At present it supports multi-core operations but does not yet exploit the new MIC SIMD instructions. We expect to release on supporting SIMD in a month or two.
There were difficulties encountered due to the absence of detailed documentation for the assembler Intel release for the chip. It was a matter of trial and error to discover its syntax. At least one bug exists in the version of the assembler we got : it generates the wrong code in one instance.
Beyond this many of the instructions that Intel introduced since the original Pentium have been deleted which caused some problems, we have had to revert for this processor to very basic IA-64 instructions. In particular there is no truncate instruction, no conditional move instructions and some restrictions on stack addressing. The absence of a truncate instruction means we have had to revert to using a subroutine for this operation.
Demonstration of the gripper functionality. The gripper is provided by flexible element which allows finger flexible moving. When the fingers touch desk of the table, the finger flexible moves back. The finger push down on the desk by force correspond to 20N / 10mm. Flexible finger allow us follow the desk of the table and grasping garment lying on the table.
The CloPeMa testbed robot arm keeps constant distance from man. The man moves in front of the robot, his position is measured by Xtion 3D range sensor and the arm is navigated to position in constant distance from recognized obstacle (man)
CloPeMa testbed robot grasping a garmetn by one gripper (arm). Man throws the piece of garment is on table, robot measures position of the piece of garment by Xtion (3D range sensor). Grasping points are detected and selected. The robot arm grasp the piece of garment using the point.