An IEEE Website
IEEE site LOGO
Stuff Center

 Home

 Branch Details

 Universal IEEE

 Engineering

 Contact

 Stuff Center
    Explorer Watch
    Unearthed
 News Letter

 Glossary

 Site Info

Site Guide :: Back to Home

Artificial Intelligence in Robots:

Robots are simply programmed to perform a series of repetitive tasks. Most of the people think that robot is a machine which performing like a human being, exhibiting a certain degree of intelligence. But robots are not capable of sensing its external environment and thinking like human being we require the direct application of artificial intelligence because all robots are microprocessor based and thus computer controlled. Robot technology has taken two Evolutionary paths : Industrial and Domestic. To perform precision assembly and control operations, industrial robots must be capable of sensing their surroundings and posses enough intelligence to respond to a changing situation. Like wise domestic robots must posses significant sensing capabilities. Thus, sensory perception and intelligence are the common denominations of any advanced robot. Sensory perception categories include vision, tactile sensing, range finding, navigation and voice communication. Artificial Intelligence means 'the science of making machines do things that would require intelligence if done by men'. Computes perform operations and make decisions using on/off states. likewise, many believe that the basic element of brain, the neuron also exhibits two state operation.

"making machines do things that would require intelligence if done by men"

The function of neuron is to generate and conduct nerve impulses. Nerve impulses to and from the neuron are propagated along fibers called synapses. These are electro-chemical impulses of approx. 50 mV with duration of about 1mS. The impulse create both chemical and physical change within the neuron body and associated fibers. Thus, the presence or absence of nerve impulses can be viewed as an on/off or two-state operation.

This excitation and inhibition or on/off effect of the synapses suggests that they operate very much like simple logic gate. The threshold level per the synapse is that of two nerve pulses are approx. 100mV it performs like a two bit AND gate. If it is 50mV then it would perform like a two bit OR gate.

The operation of the human brain is thought to exhibit a digital character at the neuron-synapse level. If this is true, the basic elements of human thought process can be linked to integrated transistor with in digital computer. Consequently, once elements of the human thought process are understood, there is no reason why computers cannot be programmed to be truly intelligent. Although the fundamental nerve-fiber action of the human brain can be described and understood. The combined effect of many nerve fibers performing an intelligent act is not fully understood. For this reason, some AI researchers study how to make computers smarter by attempting to understanding the human thought process.

Artificial Intelligence may be defined as, 'the science of making machines do things that would require intelligence if done by man'. The range of applications of a robot is enhanced most by vision. The science of providing vision for a robot is called Visual Machine Perception. Most existing vision systems employ one or more cameras to act as the robot's eyes. The features of the object such as brightness, hue and texture are transferred into digital signals that can be stored in a pixel and analyzed by a computer. Most industrial vision systems are applied to product inspection and object identification. The primary considerations for an industrial vision system are cost, speed, accuracy and reliability. These design considerations are linked to three levels of design difficulty that control the position and appearance of the object. Some industrial systems employ binary image subtraction or window and pixel counting. More sophisticated systems must employ edge and region analysis along with intelligent image-understanding techniques for object identification. Robot voice communication requires the application of speech synthesis and speech recognition. Speech synthesis is a relatively proven technology and is generally accomplished using a technique called phoneme speech synthesis. Two techniques are presently used for speech recognition : isolated word recognition and connected speech understanding.

Tactile sensing or touch for a robot is accomplished by placing micro switches, strain gauges, pressure transducers and optical sensors in the manipulators of the robot. The most important parameter that must be measured to achieve tactile sensing in fiber. Simple ranging and navigation systems employ position sensing, proximity sensing to determine the distance to an object. However, advanced systems must employ 3-D vision to navigate within the real world. Artificial Intelligence must be incorporated into intelligent robot controllers to integrate sensory information with control decisions.

To design AI into a robot involves knowledge engineering, both hardware and software design components of knowledge engineering require the networking of multiprocessor system and hierarchical planning. The software design aspect of knowledge engineering also involves the science of knowledge representation. knowledge representation can be divided into two categories such as general knowledge and expert knowledge representation. General knowledge representation requires a computer to contain a database sufficient to understand its surroundings using its various sensing abilities. Expert knowledge is extensive knowledge in a very specialized field. Expert system programs such as MYCIN gain their knowledge base by interacting with human experts.
Copyright IEEE.ORG all rights reserved.