Welcome to Scribd. Sign in or start your free trial to enjoy unlimited e-books, audiobooks & documents.Find out more
Standard view
Full view
of .
Look up keyword
Like this
0 of .
Results for:
No results containing your search query
P. 1
Research Paper on AI in Robotics

Research Paper on AI in Robotics

|Views: 28|Likes:
Published by mrunali_p
This is a research paper that was selected for presentation at a national level seminar on " ICT and its challenges for the future of India, conducted by Jai hind college",churchgate,mumbai.The focus being key researches going on and done by BARC and an overall scope of ICT for developing India
This is a research paper that was selected for presentation at a national level seminar on " ICT and its challenges for the future of India, conducted by Jai hind college",churchgate,mumbai.The focus being key researches going on and done by BARC and an overall scope of ICT for developing India

More info:

Categories:Types, Research, Science
Published by: mrunali_p on Nov 26, 2011
Copyright:Attribution Non-commercial


Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See more
See less





Application of AI in Robotics and its opportunities in India:Researches in BARC
Anup SharmaThird Year B.Sc. Student,Department of Computer Science.Vivekananda Education Society’s College of Arts,Science and Commerce.Chembur, Mumbai - 400071. India.Email:anups.mailme@gmail.com
- Artificial intelligence is a theory. Thebase object is the agent who is the "actor". It isrealized in software. Robots are manufactured ashardware. The connection between those two isthat the control of the robot is a software agentthat reads data from the sensors decides what todo next and then directs the effectors to act in thephysical world. The aim of this paper is to providebasic, background information on two emergingtechnologies: artificial intelligence (AI) androbotics and their scope in India. According to theDepartment of Trade and Industry (DTI), it isimportant to consider these emerging technologiesnow because their emergence on the market isanticipated to ‘affect almost every aspect of ourlives’ during the coming decades (DTI, 2002).Thus, a first major feature of these two disciplinesis product diversity. In addition, it is possible tocharacterize them as disruptive, enabling andinterdisciplinary.
 Keywords- AI concept, robotics, software
I. INTRODUCTIONMany researchers now feel that the goal of mimicking the human ability to solve problems andachieve goals in the real world the so-called ‘strongAI’ is neither likely nor desirable because a longseries of conceptual breakthroughs is required AIsystems are generally embedded within larger systems-applications can be found in video gamesspeech recognition, and in the ‘data mining’ businesssector. The field of robotics is closely linked to thatof AI, although definitional issues abound. ‘GivingAI motor capability’ seems a reasonable definition, but most people would not regard a cruise missile asa robot even though the navigation and controltechniques draw heavily on robotics research. AI androbotics are likely to continue to creep into our liveswithout us really noticing. Unfortunately, many of Mrunali ParabThird Year B.Sc. Student,Department of Computer Science.Vivekananda Education Society’s College of Arts,Science and Commerce.Chembur, Mumbai - 400071. India.Email:mrunal25_p@yahoo.co.inthe applications appear to be taking place amongstagencies, particularly the military that do not readilyrespond to public concern, however well articulatedor thought through.II. BASICSAI has been one of the most controversial domains of inquiry in computer science since it was first proposed in the 1950s. The ultimate aim is to makecomputer programs that are capable of solving problems and achieving goals in the world as well ashumans Today, successful AI applications range fromcustom-built expert systems to mass producedsoftware and consumer electronics. Robotics, on theother hand, may be thought of as
‘the science of extending human
motor capabilities with machines’ 
(Trevelyan, 1999). However, a closer look at thisdefinition creates a more complicated picture. For example, a cruise missile, although not intuitivelyreferred to as a robot, nevertheless incorporates manyof the navigation and control techniques explored inthe context of mobile-robotics research. This report,however, considers robotics research as the attemptto instill intelligent software with some degree of motor capability. Since many of the major areas of AI research play an essential role in work on robots,robotics will be considered here as a sub-section of AI. Many of those in industry do not use the term‘artificial intelligence’ even when their company’s products rely on some AI techniques.III. RESEARCH AREASAI, based upon the capabilities of digital computersto manipulate symbols, is probably not sufficient toachieve anything resembling true intelligence. This is because symbolic AI systems, as they are known, aredesigned and programmed rather than trained or evolved. AI software designers are beginning to teamup with cognitive psychologists and use cognitivescience concepts. Another example centers upon thework of the ‘connectionists’ who draw attention to
computer architecture, arguing that the arrangementof most symbolic AI programmers is fundamentallyincapable of exhibiting the essential characteristics of intelligence to any useful degree. As an alternative,connectionists aim to develop AI through artificialneural networks (ANNs). The emergence of ANNsreflects an underlying paradigm change within the AIresearch community and, as a result, such systemshave undeniably received much attention of late.However, regardless of their success in creatinginterest, the fact remains that ANNs have not nearly been able to replace symbolic AI. AI researchershave a variety of learning methods at their disposal.However, as alluded to above,ANNs represent one of the most promising of these. There are manyadvantages of ANNs and advances in this field willincrease their popularity. Their main value over symbolic AI systems lies in the fact that they aretrained rather than programmed: they learn to evolveto their environment, beyond the care and attention of their creator (Hsuing, 2002). Other major advantagesof ANNs lie in their ability to classify and recognize patterns and to handle abnormal input data, acharacteristic very important for systems that handlea wide range of data. As a result, they are best usedwhen the results of a model are more important thanunderstanding how the model works. To this end,these systems are often used in stock market analysis,fingerprint identification, character recognition,speech recognition, and scientific analysis of data(Stottler Henke, 2002). The following chains of reasoning, considered in isolation without supportingargument, all exhibit the Fallacy of the GiantCheesecake:• A sufficiently powerful Artificial Intelligence couldoverwhelm any human resistance and wipe outhumanity. [And the AI would decide to do so.]Therefore we should not build AI.• A sufficiently powerful AI could develop newmedical technologies capable of saving millions of human lives. [And the AI would decide to do so.]Therefore we should build AI.Once computers become cheap enough, the vastmajority of jobs will be performable by ArtificialIntelligence more easily than by humans. Asufficiently powerful AI would even be better than usat math, engineering, music, art, and all the other jobswe consider meaningful. [And the AI will decide to perform those jobs.] Thus after the invention of AI,humans will have nothing to do, and we'll starve or watch television.IV. ALGORITHMS AND GENETICPROGRAMMINGAn algorithm is defined as a ‘detailed sequence of actions to perform to accomplish some task’. One branch of algorithm theory, genetic programming, iscurrently receiving much attention. This is atechnique for getting software to solve a task by‘mating’ random programs and selecting the fittest inmillions of generations. Khan elaborates: ‘Geneticalgorithms use natural selection, mutating andcrossbreeding within a pool of sub-optimal scenarios. Better solutions liveand worse ones die – allowing the program todiscover the best option without trying every possiblecombination along the way.’V. APPLICATIONS.A. Intelligent simulation systemsThese applications are commonly used in a number of different scenarios. First, an Intelligent SimulationSystem (ISS) may be generated to learn more aboutthe behavior of an original system, when the originalsystem is not available for manipulation. Themodeling of climate systems is a good example.Second, the original system may not be available because of cost or safety reasons, or it may not be built yet and the purpose of learning about it is todesign it better. Third, an ISS might be employed for training purposes in anticipation of dangeroussituations, when the cost of real-world training is prohibitive. Such technologies are particularly welladvanced in military applications through thesimulation of war ‘games’. Another very big businessin the realm of ISSs is the videogame market,comparable to the film business in size. AI systemshave become fundamental to this industry because,unlike in film, it is often up to a computer or gameconsole to create a sense of reality for the game- player. Such standards of realism are going up all thetime.B. Intelligent information resourcesIntelligent systems must be able to provide includingvisual and audio data, in addition to commonplacestructured databases. One development in this areathat is receiving much attention is ‘data mining’, theextraction of general regularities from online data.This area is becoming increasingly important due tothe fact that all types of commercial and governmentinstitutions are now logging huge volumes of dataand require the means to optimize the use of thesevast resources.C. SensorsSensors are the perceptual interface between robots.On the one hand we have
 passive sensors
likecameras, which capture signals that are generated byother sources in the environment. On the other handwe have
active sensors
(for example sonar, radar,laser) which emit energy into the environment. Thisenergy is reflected by objects in the environment.These reflections can then be used to gather theinformation needed. Generally active sensors provide
more information than passive sensors. But they alsoconsume more power. This can lead to a problem onmobile robots which need to take their energy withthem in batteries. We have three types of sensors (nomatter whether sensors are active or passive). Theseare sensors that either record distances to objects or generate an entire image of the environment or measure a property of the robot itself. Many mobilerobots make use of 
range finders
, which measuredistance to nearby objects. A common type is thesonar sensor. Alternatives to sonar include radar andlaser. Some range sensors measure very short or verylong distances. Close-range sensors are often
tactile sensors
such as whiskers, bump panels and touch-sensitive skin. The other extreme are long-rangesensors like the Global Positioning System (GPS).The second important class of sensors is
imaging  sensors
. These are cameras that provide images of the environment that can then be analyzed usingcomputer vision and image recognition techniques.The third important class is
 proprioceptive sensors
.These inform the robot of its own state. To measurethe exact configuration of a robotic joint motors areoften equipped with shaft decoders that count therevolution of motors in small increments. Another way of measuring the state of the robot is to use forceand torque sensors. These are especially needed whenthe robot handles fragile objects or objects whoseexact shape and location is unknown. Imagine a tonrobot manipulator screwing in a light bulb.D. Effectors
are the means by which robots manipulatethe environment, move and change the shape of their  bodies. To understand the ability of a robot to interactwith the physical world we will use the abstractconcept of a
degree of 
 freedom (DOF)
. We count onedegree of freedom for each independent direction inwhich a robot, or one of its effectors can move. As anexample lets contemplate a rigid robot like anautonomous underwater vehicle (AUV). It has sixdegrees of freedom, three for its (
) location inspace and three for its angular orientation (alsoknown as yaw, roll and pitch). These DOFs definethe kinematic state of the robot. This can be extendedwith another dimension that gives the rate of changeof each kinematic dimension. This is called dynamicstate. Robots with non rigid bodies may haveadditional DOFs. For example a human wrist hasthree degrees of freedom – it can move up and down,side to side and can also rotate. Robot joints have 1,2, or 3 degrees of freedom each. Six degrees of freedom are required to place an object, such as ahand, at a particular point in a particular orientation.The manipulator shown in Figure1 has exactly sixdegrees of freedom, created by five revolute joints(R) and one prismatic joint (P). Revolute jointsgenerate rotational motion while the prismatic jointsgenerate sliding motion. If you take your arm as anexample you will notice, that it has more than sixdegrees of freedom. If you put your hand on the tableyou still have the freedom to rotate your elbow.Manipulators which have more degrees of freedomthan required to place an end effector to a targetlocation are easier to control than robots having onlythe minimum number of DOFs. Mobile robots aresomewhat special. The number of degrees of freedomdoes not need to have corresponding actuatedelements. Think of a car. It can move forward or  backward, and it can turn, giving it two DOFs. But if you describe the car’s kinematic configuration youwill notice that it is three-dimensional. On a flatsurface like a parking site you can maneuver your car to any (
) point, in any orientation. You see that thecar has 3
effective DOFs
but only 2
controllable DOFs
. We say a robot is
if it has moreeffective DOFs than controllable DOFs and
if the two numbers are the same.Holonomic robots are easier to control thannonholonomic (think of parking a car: it would bemuch easier to be able to move the car sideways). Butholonomic robots are mechanically more complex.Most manipulators and robot arms are holonomic andmost mobile robots are nonholonomic. [1]Computer scienceAI researchers have created many tools to solve themost difficult problems in computer science. Many of their inventions have been adopted by mainstreamcomputer science and are no longer considered a partof AI.FinanceBanks use artificial intelligence systems to organizeoperations, invest in stocks, and manage properties.In August 2001, robots beat humans in a simulatedfinancial trading competition. Financial institutionshave long used artificial neural network systems todetect charges or claims outside of the norm, flaggingthese for human investigation.Heavy industryRobots have become common in many industries.They are often given jobs that are considereddangerous to humans. Robots have proven effectivein jobs that are very repetitive which may lead tomistakes or accidents due to a lapse in concentrationand other jobs which humans may find degrading.Japan is the leader in using and producing robots inthe world. In 1999, 1,700,000 robots were in useworldwide. For more information, see survey aboutartificial intelligence in business.

You're Reading a Free Preview

/*********** DO NOT ALTER ANYTHING BELOW THIS LINE ! ************/ var s_code=s.t();if(s_code)document.write(s_code)//-->