woodworking height gauge

Performance for robotics and servo-mechanism

This definition means that a device only a "robot" if it contains a movable Mechanism can be influenced accessed by remote sensing, planning and actuation and control components. It does not mean that a minimum number of these components in the software must be implemented, or the "consumer changeable" Anyone who uses the device, for example, can the movement behavior have been difficult in the unit wired by the manufacturer.

Thus, the definition presented, as well as the rest of the material in this part of the book covers not only "pure" Robotics or only "intelligent" robots, but the slightly wider field of robotics and automation. These include "dumb" robots, such as: metal and woodworking machinery, "intelligent" washing machines, dishwashers and pool cleaning robots, etc. These are all examples of remote sensing, planning and control, but often not in a separate Components individually. For example, remote sensing, planning and conduct of the pool cleaning robots have been integrated into the mechanical design of the device, the intelligence of the human Developers.

Robotics is a very large extent all over the system, the realization of a task by a mechanical device pressed, an "intelligent" integration of components, many of whom are in other areas such as systems and control theory, computer science, nature Animation, engineering, computer vision, artificial intelligence, cognitive science, biomechanics, etc. In addition, the boundaries of robotics are not clearly defined, as well as its "core" ideas, concepts and algorithms in an ever-growing number of "external use" applications, and vice versa, nuclear technology (from other areas of vision, biology, cognitive science or biomechanics) are such important components always more and more advanced robotic systems.

This part of the WEBook trying to define what exactly the above-mentioned core material of the robotics domain, and to describe the structure it in a consistent and motivated. Nevertheless, the chosen structure only one of many possible "views" that one might have on the robotics domain.

At the same Direction, which is not above "definition" of robotics to be final or final, and it is used only as a rough framework for the structure of the individual Chapter

Components of robot systems

This figure shows the components that are part of all robotic systems. The purpose of this section is to describe the semantics of the terminology used to classify the chapters in the WEBook: "Remote Sensing" plan, "", "Modeling", "control", etc.

The real robot is a mechanical device (mechanism), that some moves in the environment, while physically interacts with this environment. This interaction involves the exchange of physical energy in some form or another. Both the robot mechanism and the environment is the "cause" of the physical interaction by "pressing", or experience the "impact" of interaction, through "can be measured by sensing".

Robotics as an integrated system for controlling the interaction with the physical world.

Sensory and physical activity are the openings, determined by the "controller" of the robot, the interaction of the mechanical body with the physical World. As mentioned above, the controller in an extreme, consist of only software, but in the other extreme, everything can be implemented in the hardware.

Within the controller component, several sub-activities are often referred to as:

Modeling. The input-output relationships of all control components (can be not be derived) from information that is stored in a model. This model can take many forms: analytical formulas, empirical look-up tables, fuzzy rules, neural networks, etc.

The name "model" often gives rise to heated discussions between the various research "schools" and WEBook is not interested in taking a stance in this debate: in the WEBook, "model" must be understood with its low semantics: "all information which is used to determine or influence on the input-output relationships of components in the controller. "

The other components, the bottom All models will be discussed within. A "system model" can be used to connect different components together, but it is clear that not all the robot is an System model to use. The sensing model "and" operating model "will contain the information with which to transform the raw physical data in task-dependent Information for the controller, and vice versa.

Planning. This is the activity that preceded the outcome of the possible measures selected, and the "best" one. Almost by definition, the planning can take place only on the basis of a kind of model.

Regulation. This component processes the results of the sensing and planning components to generate a set point to control. Also this scheme could work, or is not in any kind of (system) model.

The term "control" is often used instead of "regulation", but it is not unambiguously identify Domains, the use of one or the other. The importance of using WEBook clear from the context.

Scales in robotic systems

The above "components" Description of a robotic system is to be a "scale" description, ie adds the following system balances have a great influence on the specific content of the planning, remote sensing, modeling and control components at a certain scale, and consequently on the appropriate parts of the WEBook.

Mechanical scales. The quantity of the robot determined to a large extent by the limites of what happened to her. Roughly speaking, a big robot (such as is an autonomous container crane or a space shuttle) has different capabilities and control problems as a macro-robot (such as an industrial robot arm), a desktop robot (such as the "sumo" robots popular with amateur) or milli-micro-or nano-robots.

Spatial scale. There are big differences between the robot, the plot in 1D, 2D, 3D or 6D (three positions and three orientations).

Time scale. There are big differences between the robot must respond within of hours, seconds, milliseconds or microseconds.

The power density of scale. A robot must be operated to move, to drive, but need space and energy, so that the relationship between the two provides some capabilities of the robot.

System complexity scale. Complexity a robotic system to adapt to the growing number of interactions between independent sub-systems, components and control have, to this complexity.

Computational complexity scale. Robot controllers are inevitable, based on the real world, computer hardware, so they forced by the number of available calculations communication available – the available memory bandwidth and storage.

Obviously, these parameters are never completely independent of scale are on the same system. For example, a system to respond to the microsecond time can not be created by macro-mechanical, or a high number of communication interactions be associated with the subsystems.

Background Sensitivity

Finally, no description of the scientific material is still completely objective or context-free, in the sense that it is very difficult for the authors of the WEBook to 'forget' their background when writing their contribution. In this context has, robotics, roughly speaking, two faces: (i) the mathematical and technical face, the very "standardized" in the sense that that a broad consensus on the tools and theories (use the "systems theory"), and (ii) the AI face, the rather poorly standardized, not is due to a lack of interest or research efforts, but because of the inherent complexity of "intelligent behavior." The terminology and systems-thinking the two layers differ significantly from each other, so that the WEBook will include sections on the same material but written from different perspectives. This is not a "mistake" but a "feature" could, with the various views within the same WEBook only to a better mutual understanding and Lead to respect.

Research in Engineering Robotics follows a "bottom-up approach: existing and functioning systems to be extended and more flexible. Research in artificial intelligence, robotics is from top to bottom: the assumption that a series of low-level primitives is available, how could we apply them to increase the "intelligence" of the system. The boundary shifts between the two approaches are continuously ", as more and more" Intelligence is cast into algorithmic, system-theoretic form. For instance, the response of a robot sensor input was described as "intelligent behavior" in the late seventies and early eighties as well. Therefore, it belonged to AI was later shown that many sensor-based tasks could be like the surface or to optical tracking as a control To formulate problems with algorithmic solutions. From then on they have not heard AI anymore.

Robotics Technology

Most industrial robots at least the following five parts:

Sensors, Effectors, actuators, controllers and shared effectors known as weapons.

Many other robots via artificial intelligence and effectors to help achieve it mobility.

This section describes the basic technology of the robot. Click a link above or use The navigation menu on the right side.

Robotics Technology – Sensors

Most robots today are almost deaf and blind. Sensors can to a limited feedback to give the robot so that it can do its work. For the purposes of skills and even the simplest living organisms compared Robots have to go a very long way.

The sensor sends information to the form of electronic signals back to cfontroller. Sensors give the Robot control information about its surroundings and lets them know about the exact position of the arm or the state of the world around him.

Seeing, hearing, touch, Smell and taste are the types of information that we in our world. Robots can be designed and programmed so that one specific information beyond what is can our 5 senses will tell us. Thus, for example, a robot-sensor "sees" in the dark, detect tiny amounts of invisible radiation or measure movement is to see small or fast for the human eye.

Here are some things to sensors are used for:

Physical Properties

Technology

Contact Bump, Switch

Distance ultrasound, radar, infrared

Light Level Photocells, cameras

Sound Microphones

Strain DMS

Rotary Encoder

Magnetic compasses

Chemical Odor

Temperature heat, infrared —

Inclinometers inclinometers, gyroscope

Pressure gauge

Altitude Altimeter

The sensors can be stored on simple and complex, will be made dependent on how much information needs to be. A switch is a simple on / off sensor for turning the Robots are on and off. A human retina using a complex sensor systems) more than one hundred million light-sensitive elements (rods and cones. Sensors provide information to the robot brain, which can be treated in different ways. For example, we can easily respond to the sensor output: When the switch opens is when the switch is closed, go.

Stages of processing

To determine whether the Switch is opened or closed, you need the voltage goes through the circuit to measure the electronics. Now you can say that you have a microphone and you want to recognize a voice and they are separated from the noise that the signal processing. Now you can have a camera, and you want the image previously processed and are now You find out what the objects, perhaps by a comparison with a large library of drawings, the calculation is made. Sensory processing is a very complex matter in order to try, but not the robot needs this in order to have a "brain." The brain has to have analog or digital processing capabilities to everything connecting cables, the support electronics go to the computer and batteries to power supply for the whole thing to handle the sensitive data. Perception requires the robot sensors (electricity and electronics), calculation (more power and electronics, and (connections) to connect them all.

Switch Sensors

Position Sensors are the easiest of all. They work without processing to the electronic circuit () level. The general rationale is that an open vs. closed system. If a switch is open, no current can flow when it is closed, current flow and can be detected. This simple Principle can (and is) used in a variety of ways.

Switch sensors can be used in a variety of ways:

Contact sensors detect: if the sensor has been contacted by another object (eg, triggered when a robot encounters a wall or reaches an obstacle, they can even whiskers)

Sensors detect limit: if a mechanism that is the end of its range moved

Encoder sensors: determines how often a shaft rotates by clicking a button (open / close) every time it rotates (eg, the trigger for each round, so for the counting of the rotations)

There are many common Switches: Button switch, mouse switches, key board keys, cell phone keys, and others. Depending on how a switch is wired, it can be NO or NC. This would, of course, robots on your electronics, mechanics down, and his task. The simplest, but very useful for a robot-sensor is a "bump-switch" that tells him when he came in something, so it can again and again removed. Even such a simple idea, there are many different ways of implementation.

Light Sensors

Switch Measure physical contact and light sensors measure the amount of light effects a photoelectric cell, which basically a sensor. The resistance of a photocell is low when bright lighted, ie, when it very easily, it is high when it is dark. In this sense, a light sensor is really a "dark" sensor. When setting up a photocell Sensor is you end up with the help of the equations, we have learned about, because you need to familiarize yourself with the ratio of the photo cell photo much resistance, and resistance and the tension in the electronic sensor circuit. Of course, since you will measure the electronic structure and the writing of the program and use the output of light sensor, can To edit any time, so it's easier and more intuitive. It encloses a light sensor to its properties. The sensor can be shielded and positioned in various ways. Multiple sensors and they can be arranged from each other with shields into useful configurations.

But how can switch light sensors in many different ways be used:

Light sensors can measure:

Light intensity (how bright / dark) is

Differential intensity (difference between the photo cells)

Break-Beam (change / decrease in intensity)

Light sensors can be shielded and focused on different ways

Your position and directionality on a robot can make a big Difference and the impact

Polarized light

"Normal" light from a Source is not polarized, that is, it runs at all orientations in relation to the horizon. However, if it is a polarizing filter in front of a light source, only the light waves pass through the filter of a particular orientation. This is useful because now we are changing these remaining light with other filters can, if we send it through another filter with the same feature level, almost all of it through. But if we are) a vertical filter (a characteristic with a 90-degree angle relative to be we block all the light. Polarized light can be used to make specialized sensors from simple photo cell if you set a filter in front of a light source and the same or any other filter in front of a photocell, you can manipulate sent to identify what and how much light it.

Resistive Position Sensors

We have said earlier, a photoelectric sensor with a resistance device. We can also sense resistor in response to other physical properties, such as bending. The resistance of the device increases with the amount that is bent to them. This curve sensors were originally designed for video game control developed (eg, Nintendo Powerglove) and have tended to be quite useful. Note that you will wear and repeated bending of the sensor. No wonder, a bend sensor is much less robust than light sensors, although they share the same basic resistive principle.

Potentiometers

These devices are very common for manual tuning, you've probably seen them in some controls (such as volume and tone) in – stereo systems. usually called pots, allowing them the user to manually adjust the resistance. The general idea is that the device has a movable tap on two fixed ends there. When the valve moves, the resistance changes themselves. As you can imagine, the fixed resistor between the two ends is, however, the resistance between the movable part and the two ends differ, as part of will be postponed. In robotics, pots commonly used sense and tune position for sliding and rotating mechanisms.

Biological Analogues

All sensors are described in biological systems, we

Touch / contact sensors with much precision and complexity in all types

Bend / Resistance receptors in the muscles

Reflective Sensor Accessories

We mentioned that if we can change a light bulb in combination with a photodetector, we take a break beam sensor. This idea is based on the principle of the reflective optical sensors: The sensor consists of a transmitter and a detector. Depending on the arrangement of these two together, we can have two types of sensors:

Diffuse sensors (Transmitter and detector next to each other, separated by a barrier, the objects are detected when the light reflected from them and back into the detector)

Break-beam (emitter and detector facing each other, objects are detected when the beam between the emitter and detector Interrupt)

The transmitter is usually made out of a light-emitting diode (LED on), and the detector is typically a photodiode / phototransistor.

Note that this is not the same technology as resistive photocells. Resistive photocells are nice and simple, but make their resistive properties slowly, photodiodes and photo transistors are much faster and therefore the preferred type of technology.

What can you with this simple idea to do the reflection of light? A lot of useful things:

Object presence detection

Object distance detection

Surface feature detection (detection / next marker / tape)

The wall, tracking boundary

Axis Encoding (with encoder wheels with bars or black and white color)

Bar code decoding

Note, however, depends depends on that the light reflection on the color (and other properties) with a surface. A bright surface reflects light better than dark, and a black Surface may not reflect it at all, it seems invisible for a light sensor. Therefore, it may be harder (less reliable) to dark objects recognized in this way as a lighter. In the case of object distance, lighter objects that are farther away will seem closer than darker objects that are not so far away. This gives You an idea of how the physical world is to observe partially. Even if we are useful sensors, we have not fully complete and accurate information.

Another source of noise in light sensors is ambient light. The best thing to do is subtract the ambient light level of the sensor reading to the actual Change of the reflected light, not the ambient light detection. How does it work? The two (or more, for higher accuracy) readings of the detector, with a transmitter on, and with it off, and subtracting the two values from each other. The result is the ambient light, which can then be subtracted from future readings. This process is called sensor calibration. Of course, remember that the ambient light levels change, so the sensors must be able to be calibrated again and again.

Break-beam sensors

We already talked about the idea of the break-beam sensors. In general, each Pair of compatible emitter-detector devices can be used to produce such sensors:

a light bulb, flashlight lamps and a photocell

red LEDs and visible light sensitive photo-transistors

Infrared or IR emitters and detectors

Shaft Encoding

Encoders measure the rotation angle of an axis providing position and / or the speed info. For Sample is measured by a tachometer to address how quickly the wheels of a vehicle, while an odometer, the number of revolutions of the wheels taken.

To recognize the mark a complete or partial rotation, we need some way to the rotary element. This is usually done by attaching a circular disk on the Shaft and cut notches into it. A light transmitter and receiver are located on each side of the plate are such that the groove runs between them, the light and is detected where there is no notch in the disc, is no light.

If it is just a notch in the disk, then a rotation is detected, as it happens. This is not a very good idea, because it is only a low level of resolution for measuring speed: the smallest unit that can be measured is a full revolution. In addition, some might Rotations will be missed by noise.

In general, many notches are cut into the plate, and the effects of light hits the detector are counted. (You can see that it is important to have a fast sensor here, when the shaft rotates very quickly.)

An alternative to cutting notches absorbing in the drive to the hard disk with a black (not reflective paint) and white (highly reflective) wedges, and measure the reflectance. In this case, the emitter and detector on the same side of the disc.

In any case, the output of the sensor is going to a wave function of light intensity. These can then develop methods to measure the speed, by counting the peaks of the waves.

Note that the wave action, at both encoding Position and speed of rotation, by subtracting the difference in the position, readings after each time interval. Speed, on the other hand, tell us how quickly a robot moves, or if it moves at all. There are several possibilities for using this measure:

measure the speed of a powered (active) Cycling

Using a passive wheel, pulled by the robot (action will continue to progress)

We combine the Position and velocity information to do in order to more complicated things:

move in a straight line

rotated by You an exact amount

Note, however, that to do such things quite difficult, because the wheels (effector noise and errors) and slide They slide and tilt it is usually some slop and backlash in the gearing mechanism. Encoder can provide the feedback to correct the error, but because some To avoid mistakes.

Quadrature Shaft Encoding

Previously, we reported the detection of position and velocity spoken, but not talk about the direction of rotation. Suppose the wheel suddenly changes the direction of rotation, it would be useful for the robot to recognize, dass

An example of a common system for the needs of position, velocity measurement, and the direction is a computer mouse. Without certain direction, a mouse is pretty useless. As the rotation is measured?

Quadrature shaft encoding is an elaboration of the fundamental break-beam Idea, instead of using only one sensor is needed two. The encoders are aligned so that the two data streams coming from the detector and fourth cycle (90 degrees) from the Phase, hence the name "quadrature". By comparing the performance of the two donors at each time step with the output of the previous time step, we can say that if to change a direction. When the two at each time step the sample, only one of them to change its state (ie have to go from on to off to) at a time when they made are phase. Which one does it determine which direction the shaft rotates. Whenever a wave is moving in a direction that increases a counter, and if it is in the opposite Direction, the counter is decremented, so that the pursuit of the general situation.

Other uses of quadrature shaft encoding in robot arms with complex Joints (eg, rotational ball joints, remember) on your knee or shoulder), Cartesian robot (and big printer, where an arm / rack moves back and forth along an axis / gearbox.

Modulation and demodulation of light

We mentioned the ambient light is a problem because they Problems with the light emitted by a light sensor. One way around this problem is to emit modulated light, ie, quickly turn the transmitter on and off. Such a signal is much easier and more reliably detected by a demodulator that is tailored to the specific frequency of the modulated light. It is not surprising need a detector to several on-flashes in a series of sense to detect a signal, ie, to recognize their frequency. This is a small point, but it is in writing code demodulator important.

The idea of modulated infrared light is often used, for example, in the household remote controls.

Modulated light sensors are generally more reliable than basic light sensors. They can be used for the same purposes: to demonstrate the presence of an object measuring the distance to a nearby object (smart electronics is required, of course, see notes)

Infrared (IR) sensors

Infrared sensors are a type of light sensors, which function in the infrared part of the spectrum. Infrared sensors exist that are active sensors: They consist of a transmitter and a receiver. IR sensors are in the same way that visible light sensors that we used previously discussed: as a break-beam and as a reflection sensors. IR is preferable for visible light in robotics (and other) applications, which is suffering a little bit less of the surrounding faults, because they are easily modulated, and because it is simply not visible.

IR communication

Modulated infrared can be used as a serial interface for transmission of messages be used. This is is a fact, work as IR modem. Two basic methods exist:

Bit part (in the middle of each bit is sampled, takes take all the bits transmitted the same time)

Bit intervals (more common in commercial use; sampled at the falling edge, the Intended duration of the interval between sampling, whether it is a 0 or 1)

Ultrasonic Distance Sensing

As we mentioned before, ultrasonic sensor on the time-of-flight principle. The transmitter produces a sonar "chirp" of sound that travels from the source, and if it encounters obstacles, reflecting on them and) returns to the receiver (microphone. The amount of time, the return for the proper beam tracked (is using a timer, when the "chirp is produced," and she stops when there, the reflected sound), and is used to traveling the distance of the sound calculated. This is possible (and easy), because we know how fast the sound, which is a constant which varies slightly based on ambient temperature.

At room temperature, the sound up to 1.12 meters per millisecond. Another way to get it that the sound in 0.89 milliseconds per foot. This recall is a useful constant.

The process of finding a place on the basis of sonar called echolocation. The inspiration for Ultrasonic Sensor comes from nature; Bats using ultrasound instead of vision (this makes sense, they live in very dark caves, where vision would be) largely useless. Bat sonar systems are extremely demanding in comparison to artificial sonar, used it for many different frequencies in the search for even the smallest fast-flying prey and to avoid hundreds of other bats, and communication for the search for partners.

Specular Reflection

A major disadvantage of Ultrasonic sensor) is its vulnerability to specular reflection (mirror reflection means on the outer surface of the object. While the sonar-based sensor on the sound wave of reflective surfaces and the return is based on the recipient, it is important to remember that the sound waves do not necessarily bounce off the surface and "come right back." In fact, the direction of reflection depends on the angle of the sound beam and the surface. The smaller the angle, the higher the probability that the sound only graze "on the surface and bounce off, not back into the emitter that produces again a false long / distant reading. This is often seen as a reflection, as inclined smooth surfaces with reflective qualities to worsen this reflection problem. Rough surfaces produce more irregular reflections, some of which go back more to the emitter. (For example, in our robotics lab on campus, we use sonar sensors, and we have a portion of the test-lined box, because it is much better sonar reflective properties than the very smooth wall behind him.)

Summary it must be held long sonar readings can be very inaccurate, since they may reflect the result of false and not accurate. This must be taken into account when programming manufacture of robots or a robot that can be very undesirable and unsafe behavior. For example, a robot may be approaching a wall at a steep angle does not see the Wall at all, and collide with him!

However, the sonar sensors have been successfully built for very demanding robotic applications, including the terrain and indoor mapping is used, and remain a very popular choice of sensor in mobile robotics.

The first commercial ultrasonic sensor was produced by Polaroid and used (in order to automatically measure the distance to the nearest object is probably the most photographed at present). These simple Polaroid sensors still too the most popular off-the-shelf sonar systems (they come with a processor board that is) with the analog electronics. Their standard features include:

32-foot range

30-degree angle

Sensitivity to specular reflection

shortest Distance back

Polaroid sensors can be combined into phased arrays to more complex and more accurate sensors.

Ultrasound can be found in a variety of other applications, the best known is between U-boats. The sonars there was much more focused and have longer-range Beams. Simpler and more mundane applications include automated tape-measures ", height measures, alarms, etc.

Machine Vision

So far we have spoken of relatively simple sensors. They were simple in terms of processing the information it back. Now we turn to machine vision, ie, the cameras as sensors.

Cameras, of course, model biological eyes. Needless to say, all biological eyes are more complex than any camera we know today, but as you will see the cameras and machine vision systems, the processing of perceptual information, not easy at all! In fact, the machine vision is such a challenging topic that it has long been a separate department for Artificial Intelligence.

The general principle of the camera is that light from objects in the distributed environment () to read the scene that passes through an opening ( "Iris" In the simplest case, a pin hole in the more advanced case, a lens) and striking the so-called image plane. In biological systems, the image plane of the retina that is too many Rods and cones is attached (light-sensitive elements) which are in turn to perform nerves, so-called "early vision of" connected, and you will then pass the information on the entire brain to "higher level" vision processing. As we mentioned earlier, is a very large percentage of people (and other animal) brain devoted to visual processing, so this is a very complex exercise.

In cameras, instead of light-sensitive Rhodopsin rods and cones, we use silver halides in photographic film or silicon circuits) with CCD sensors (CCD cameras. In all cases, some information ) on the incident light (eg color intensity of these light-sensitive elements detected on the image plane.

In machine vision, must The computer sense is made of the information is on the image plane. When the camera is very simple, and uses a tiny pin hole, then some calculation is necessary in order to Calculate the projection of the objects in the environment on the image plane (note, it is inverted). When a lens is involved (as in vertebrate eyes and) real cameras, you can then more light in, but is concentrated at the price, only objects of a certain number of lines from the lens be in focus. This area is recognized as the distances of the Camera depth of field.

The image plane is usually in equal parts, called pixels, typically arranged in a rectangular grid subdivided. In a typical camera has 512 by 512 pixels on the image plane (for comparison, there are 120 x 10 ^ 6 poles and 6 x 10 ^ 6 cones in the eye, arranged in hexagonal). We call the projection onto the image plane of the image.

The brightness of each pixel of the image is proportional to the amount of light into the camera under the direction of the surface patch of the object that the projects in this pixel. (This of course depends on the reflective properties of the surface patch, the location and distribution of light sources in the Environment, and the amount of light from other objects in the scene on the expression of surface patch.) As it turns out, a brightness of a patch, depends on two types of Considerations, one of which (from the surface, reflection, as we saw before), and the other is diffuse (light penetrates into the object, is absorbed and then re-emitted). Reconstruct the correct model for light reflection, and the scene, all these properties is necessary.

Let us assume that we are dealing with a black and white camera with 512 x 512 pixel image plane. Now we have an image that is a collection of pixels, each having an intensity between white and black. To find an object in this picture (if there is one, of course, we do not know a priori) know, the typical first step ( "early vision") is to do edge detection, ie, all Edges. How do we recognize them? We define edges as curves in the image plane where there are significant changes in brightness.

A simple approach would be to search for sharp changes in brightness by the staggering of the image, and to areas where the size of the derivative is considered large. This almost works But unfortunately, all kinds of false peaks, which produces noise. We also can not be distinguished from natural to changes in intensity through the shadow of which result physical objects. But let's leave that now and think about noise. How do we deal with noise?

We smoothing, ie, a mathematical procedure, we considered to be folding, and finds that eliminates the isolated peaks. Folding, in fact, applies a filter to the image. In fact, to find any edges in the image, we need to Picture with many filters convolve with different orientations. Fortunately, the relatively complicated mathematics involved in edge detection has been well studied, and now there are standard and preferred approaches the edge detection.

Once we have edges, is the next thing to do to try to find objects at all edges. Segmentation is the process of dividing up or organize the image into parts correspond to the continuous objects. But how do we know which lines correspond to which objects, and what makes an object? There are several indications we can use to detect objects:

We can model the line-drawings of objects stored have (from many possible angles, and possible at many different scales!), and then compare them with all possible combinations of edges in the picture. Note that this is a very compute-intensive and expensive. This general approach, which has extensively studied, is a model-based vision.

We can take advantage of the movement. If we get a picture of two consecutive time-steps, and we are moving between the camera in any continuous solid objects (which obeys the laws of physics) is a, that is moved, its brightness will remain properties. These hives us a clue to search for objects by Subtraction of two images of each other. Note however that this also depends on, knowing how the camera relative to the scene (direction, distance), postponed and nothing moved in the scene at the time. This general approach, which has also been extensively studied, is called motion vision.

We can Stereo (ie, binocular stereopsis, two eyes / cameras / points of view). Just as with motion vision above, but without actually moving, we get two images that we see each other can subtract when we will know what should be differences between them, that is, if we know how the two cameras are organized / position to each other.

We can texture. Patches have that uniform texture is consistent, and have almost identical brightness, we can assume that they come from the same object. Because we are able to extract a statement, which may include parts on the same object in the scene get.

We can also provide shade and contours in a similar manner. And there are many other ways in object shape and projective invariants, etc.

Note that all of the above Strategies are employed in biological vision. It is difficult to unexpected places, or completely new methods to recognize (because we do not have no models or not) at the ready. Movement helps to catch our attention. Stereo, ie, two eyes, is of crucial importance, and use all the meat-eaters (they have two eyes in the same direction, in contrast ) to herbivores. The brain has an excellent job of quickly extracting the information that we need for the scene.

Machine Vision to do the same task real-time vision. But that, as we have seen, a very difficult task. Often try an alternative, to do all of the above steps to object recognition do, it is possible to simplify the vision problem in various ways:

Use color to look special and unique colored objects to recognize them and in this way (such as stop signs, for example)

Use a small image plane, but a full 512 x 512 pixel array we can reduce our view, much less, because (for example, only one row is called), this is a linear CCD. Of course there is much less information in the image But if we are smart and know what to expect, we can handle what we see is useful and fast.

Use other, simpler and faster, sensors, and combine those with a vision. For example, isolate, IR cameras, people, body temperature. Grippers allow us to touch and objects move by which we can be sure they exist.

Use the information on the environment when you know you will be on the street, the white lines, looks specifically for these lines in the right places in the picture. This is, first and still the fastest road and highway Robots will take place.

These and many other sophisticated techniques are used when one considers how important it is to "see" in real time. Consider the highway driving as an important and growing application of robotics and AI. Everything moves so fast that the system must recognize and act in order to the time to react protectively and securely, as well as intelligent.

Now that you understand the complexity of vision, you can see why it is not on the first robot was used, and it is still not to be used for all applications, and definitely not a simple robot. A robot can be very useful, without Vision, but some jobs require it. As always, it is important to think about the proper coordination between the sensors of the robot and the task.

About the Author

Assistant professor in lord venkateswara engineering college.I am doing phd in sathyabama university, Tamil Nadu,India.

scmi Logic 23 5 head moulder at Pleasant St Woodworking Machinery BRLTB

Since you found this site, you must enjoy working with wood. And if you’re like me, you’ve probably spent hundreds of hard-earned dollars on wood plans and projects that turned out to be disappointing (or maybe even disastrous!). Maybe you’ve halted projects in mid-stream because of frustration or from running out of ideas. To read more, please click here:Woodworking4Home