Central Computing Facility
Since its inception, Computer Division has been providing state-of-the art computing infrastructure and services to the scientists and engineers of the Centre. At present one two-node ES-40 Alpha server with aggregate computational power of 6 GFLOPS is in operation since 2001. The operating system is COMPAQ Tru 64 UNIX Ver 5.1 with Tru Cluster. The facility has been augmented with the installation and commissioning of one two-partition rx7640 Itanium server with aggregate computational power of 96 GFLOPS and a total (raw) disk capacity of 4.8 TB. Software consists of 64-bit HP-UX and RHEL-AS4 for IPF operating system with various compilers, libraries and development tools.
IT and Grid Activities
Computer Division, VECC is engaged in providing IT Services in general and Web Services and Indico Services, in particular. The homepage of VECC is hosted on a load-balanced, security-enhanced architecture to guard against cyber threats. All IT-services are run on a load-balanced redundant platform. The Indico services include management of Seminars, Conferences, Meetings and the like. Various IT applications (like Database service and Ticketing Service) have been developed in-house to cater to the requirements of users.
The Computer Division has been engaged in Grid-related activities since early 2000 by way of contributing to the development of Grid Middleware for the Large Hadron Collider Computing Grid (LCG). Support for maintaining the Kolkata Tier-II Centre for the ALICE experiment was extended till 2009. The proposal put forward by VECC in 2004 for setting up an intra-DAE grid, encompassing a few R&D units of DAE, is now a reality.
GARUDA, which stands for Global Access to Resources Using Distributed Architecture, is India’s first national grid computing initiative that aims to provide the e-Infrastructure required to enable data and compute intensive science for the 21st century in the country. GARUDA grid is an aggregation of resources comprising of computational nodes, mass storage and scientific instruments and hosts scientific applications through the National Knowledge Network (NKN). The NKN is a state-of-the-art multi-gigabit pan-India network for providing a unified high speed network backbone for all knowledge related institutions. VECC is a Partner Institution of the GARUDA grid and has access to the GARUDA resources through the 1Gbps NKN connectivity.
VECC has participated in the EU-India Grid Project during 2007-2009 and contributed towards development of High Energy Physics applications on the Grid. It has developed, and is currently supporting a test-bed for the EU-India Grid partners. It was also a member of the EU-India-Grid2 project that started on 1st January, 2010 and ended on 31st December, 2011. Another ambitious project, The Health Grid Initiative, for setting up a collaborative framework for the development of health-care and medical applications with the latest grid computing technology is currently underway.
Educational Software in Sign Language for the Hearing Impaired
The hearing impaired community has traditionally found itself marginalized from the mainstream of the society. The inability of the hearing impaired persons to communicate orally, coupled with the fact that traditional education has cared little to address their special requirements and needs, has come in the way of their self-development. They have, thus, found themselves in a disadvantageous position in almost all spheres of life. Survey reveals that around 90% of the world’s deaf children and adults have never been to schools and are thus more or less illiterate. Hence, there arises the need to develop tools and appliances, which would not only help them in communicating with the hearing mainstream of the society, but would also enable them to embrace basic education easily.
With an estimated deaf population of several millions, India can ill-afford to remain indifferent to the needs and aspirations of this segment of the society. It is thus incumbent upon the technical community of India to provide before the hearing impaired community, aids and appliances, which would help mitigate, if not demolish, the barrier they face while communicating with the normal people in the society.
The Department of Atomic Energy has always been appreciative of the need for development of technology for the betterment of life of the citizens of this country. With this mandate of the Department in the backdrop and realizing the hardships faced by the hearing-impaired persons, the Computer & Informatics Group, Kolkata, in collaboration with Webel Mediatronics Ltd, Kolkata, has developed, in the XI plan period, software for automatic conversion of text to sign language. The software, named Mounisara, not only helps mainstream the alienated community of the hearing impaired, but also is an effective educational aid for the deaf children.
Sign language is a common means of communication amongst people suffering from hearing impairment. It takes recourse to visual transmission of gestures to convey meaningful information. A gesture in sign language is the counterpart of a word in an oral language. Hence, the equivalent of a sentence in an oral language is a sequence of gestures in sign language. Sign languages have their own grammar and are not just direct translations of the spoken words into gestures.
When a sentence (or word) is typed into Mounisara, it generates a visual display on the computer screen of the gestures suggestive of the thought contained in the sentence (or word) typed in. Thus, Mounisara acts as a translator from text to sign language. The developed translator-software provides an intermediary for communication between the hearing-impaired persons and the normal people. Since an input text gets displayed on the screen as a sequence of gestures, one-way communication with the deaf becomes easily possible.
Also, persons desirous of learning the sign language (for example, friends, relatives, teachers of deaf children) can take the help of this translator. They can input words and sentences into the software and get the corresponding sign language output displayed on the computer screen. By repeated viewing and practice, they can familiarize themselves with the sign language. It is particularly beneficial for the teachers who have deaf children as their students.
The first version of Mounisara was formally released on June 08, 2009 by Director, VECC in presence of the Dr. Debesh Das, the then Hon’ble Minister-in-Charge, Department of Information Technology and Biotechnology, Government of West Bengal.
|
| Graphical user interface of Mounisara 1.0. Formal release of the software at VECC on 8/06/2009. |
This software is being beneficially used as an educational aid in several schools for the deaf children in West Bengal. This newspaper report carries comments of the teachers of two schools, who reveal how this software is benefitting the deaf children.
In addition, we have also developed e-books for the primary school children based on sign language. These e-books embody the contents of standard text books of primary school children in sign language, with textual subtitle. So far we have developed e-books for the Bangla text books Barnaporichoy (Part I and Part II), Kishalaya (for class I) and Sahaj Patha (for class I). Lessons on conversation in sign language are also available in software.
 |
 |
| Graphical user interface of e-book (Text Book: Barnoparichoy – II). | Graphical user interface of conversation lessons. |
Buoyed up by the appreciation received for our developments, we have developed similar translator software where the input text is Hindi.
 |
| Graphical user interface of the Hindi version of Mounisara 1.0. |
These softwarea are freely available on DVDs to all institutes and organizations associated with the hearing-impaired persons. For a copy, a written request may be sent to the Head of the Computer & Informatics Group of this Centre.
[1] B. Sarkar, K. Datta, C. D. Datta, D. Sarkar , S. J. Dutta, I. Das Roy, A. Paul, J. U. Molla, A. Paul,“A Translator for Bangla Text to Sign Language”, Annual IEEE India Conference – 2009 (IEEE INDICON – 2009), Gandhi Nagar, December 18 – 20, 2009
[2] Pathik Guha, “Sign language for education of the deaf and dumb is now also through software” [in Bangla], Anandabazar Patrika, December 03, 2011.
[3] Mathures Paul, “End of silence”, The Statesman, April 29, 2012.
Localization of Mobile Robots in Indoor Environments
A mobile robot when deployed for some applications like material transfer, must have the capability of navigating autonomously in its environment. One of the pre-requisites for autonomous navigation is the capability of self-localization (or simply, localization), which continually provides the robot with an answer to the question “Where am I?” In other words, it enables the robot to reliably estimate its location and orientation (heading direction) in its environment with respect to an external coordinate system.
If the initial pose (position and orientation) of a robot is known, one way of estimating the subsequent poses is through odometry. Odometers are simply encoders attached to the wheels of the robot for measuring revolutions of the robot’s wheels. By continually integrating the readings of the encoders to the initial pose, an estimate of the pose at any instant of time may be made. However, this estimate soon becomes unreliable because of errors that grow without bound. But, if the locations of the objects in the vicinity of the robot are accurately known, and the robot is able to accurately perceive its distance from these objects, it is possible to make a reasonably accurate estimate of its pose. Thus, we need to have an accurate map of the environment of the robot in the first place. A map is essentially a model, or an abstraction, of the environment described with respect to some world coordinate system. All objects in the environment find a representation in the map in one form or the other, depending upon the type of map. Once an accurate map of the environment is available with the robot, it only needs to estimate its distance from a few nearby landmarks and from there figure out its pose.
We have formulated and implemented a method to extract line map from consistently registered 2D laser range data of a polygonal environment. The figure below shows a line map obtained using the implemented method. A SICK LMS-200 laser range finder on a Pioneer-3DX mobile robot was used to acquire the range data.
We have also developed methods for localisation in office-like indoor environments using scan matching and particle filters. These methods use line maps and laser range scan data to achieve localization.
Assessment of Radiation Risk in Cyclotron-Adjoining Areas Using Mobile Robot
The K-130 Room Temperature Cyclotron at Variable Energy Cyclotron Centre (VECC), Kolkata delivers accelerated proton, alpha and other positive heavy ion beams of different energies. High-level mixed-radiation fields of neutrons and gamma rays are produced during the operation of the cyclotron. Persons, associated with the operation and maintenance of the machine, have to be protected from the radiation exposure in normal and in accidental scenarios. Hence, the estimation of radiation dose rates at various locations around the machine inside the active areas, such as vault and pit (inaccessible areas during operation), and in the experimental areas is essential. A mobile robot based radiation dose-rate mapping system has been developed to estimate the distribution of the radiation field prevailing during the operation of the Cyclotron at different regions inside the Cyclotron vault and the experimental areas.
The mobile robot is tele-operated from a computer located in the Cyclotron control room over wireless-Ethernet and its position and orientation are monitored with the help of five cameras mounted at different positions inside the cyclotron vault. The mobile robot moves around the Cyclotron and its adjoining areas while beams are delivered. It measures the radiation dose rates and communicates the measured values to the remote computer for visualization.
[1] M. Purkait, S.C. Jena, T.K. Bhaumik, K. Datta, B. Sarkar, C. Datta, D. Sarkar, R. Ravishankar, S. K Mishra, T. Bandyopadhyay, S. Sharma, V. Agashe, P. K. Pal, “Online Radiation Mapping of K-130 Room Temperature Cyclotron using Mobile Robot”, Proceedings of 2nd IEEE International Conference on Computer and Communication Technology – 2011, Allahabad, India, 15 – 17 September, 2011 pp. 104 - 106.
Research on Image Processing and Soft Computing Methods
Research work in the areas of image processing and soft computing methods, leading to publications in international journals, has been carried out. In the following we give a brief account of some of our activities in these areas.
Several different formulations for approximation of digitized curves, which find applications in image analysis and pattern recognition, have been developed. A method for polygonal approximation of digitized curves using the geometric properties of the curve wasformulated. A measure for assessing the goodness of approximation (‘Figure-of-Merit’) was also proposed. Later, methods based on genetic algorithm and differential evolution, for polygonal approximations were proposed. A method for approximation of digitized curves using circular arcs and straight line segments using genetic algorithm was also developed. A new method for hierarchical representation of digitized curves was also proposed.
A scheme for representing binary images based on minimization of Boolean functions has been developed. The binary image is considered as a map of Boolean function and the Quine-McCluskey method has been applied to minimize the function. The essential prime implicants represent the image in a set-of-codes form. Drastic saving of storage is achieved compared to linear quadtrees (LQ) and interpolation-based bintrees (IBB). Algorithms of various set operations (intersection, union, complement), geometric operations (area, centroid), translation, rotation,mirroring, connected components labeling and topological properties of images encoded as minimized Boolean function (MBF) have been developed. A generalized approach for finding boundary codes from region representation of set-of-codes type has been proposed. Algorithms for operations on binary images represented by IBB have also been developed.
The notion of a concavoconvex fuzzy set in the Euclidean space has been introduced by combining the concepts of convex and concave fuzzy sets. Useful properties of the concavoconvex fuzzy set have also been established.
Partial List of Journal Publications
• B. Sarkar, “An efficient method for near-optimal polygonal approximation based on differential evolution”, International Journal of Pattern Recognition & Artificial Intelligence, Volume 22, Number 6, 2008, pp. 1267 – 1281.
• B. Sarkar, L. K. Singh and D. Sarkar, “A genetic algorithm-based approach for detection of significant vertices for polygonal approximation of digital curves”, International Journal of Image and Graphics , Volume 4, Number 2, 2004, pp. 223 –239.
• B. Sarkar, S. Roy and D. Sarkar, “Hierarchical representation of digitized curves through dominant point detection”, Pattern Recognition Letters, Volume 24, Issue 15, November 2003, pp. 2869 – 2882.
• B. Sarkar, L. K. Singh and D. Sarkar, “Approximation of digital curves with line segments and circular arcs using genetic algorithms”, Pattern Recognition Letters, Volume 24, Issue 15, November 2003, pp. 2585 –2595.
• D. Sarkar and P.K. Das, “On mirroring, connected component labelling and topological properties of images encoded as Minimized Boolean Function”, Informatica, Volume 26, Number 1, 2002, pp. 1267 – 1281.
• P. K. Pal and B. Sarkar, “A language to model animation out of behaviour-embedded graphical components”, The Journal of Visualization and Computer Animation, Volume 13, Issue 3, 2002, pp.169 – 185.
• B. Sarkar, L. K. Singh and D. Sarkar, “Approximation of digital curves with line segments and circular arcs using genetic algorithms”, Pattern Recognition Letters, Volume 24, Issue 15, November 2003, pp. 2585 –2595.
• D. Sarkar and P. K. Das, “Generalized approach for finding boundary codes from region representation of set –of-codes type”, International Journal of Pattern Recognition & Artificial Intelligence , Volume 14, Number 8, 2000, pp. 1039 – 1052.
• D. Sarkar and N. Gupta, “Operations on binary images represented by interpolation based bintrees”, Pattern Recognition Letters, Volume 20, Number 4, 1999, pp. 395-403.
• D. Sarkar, “Operations on binary images encoded as minimized boolean functions”, Pattern Recognition Letters Letters , Volume 18, 1997, pp. 455-463.
• D. Sarkar,S. Banerjee and S. Chattopadhyay, “Translation and rotation of binary images encoded as minimized boolean functions”, Pattern Recognition Letters, Vol. 18, 1997, pp.157-163.
• D. Sarkar, “Boolean function-based approach for encoding of binary images”, Pattern Recognition Letters, Volume 17, 1996, pp. 839-848.
• D. Sarkar, “Concavoconvex fuzzy set”, Fuzzy Sets and Systems, Volume 79, Number 2, 1996, pp. 267-269.
• D. Sarkar, “A simple algorithm for detection of significant vertices for polygonal approximation of chain-coded curves”, Pattern Recognition Letters, Volume 14, Number 12, 1993, 959-964.