Docstoc

wireless robotics

Document Sample
wireless robotics Powered By Docstoc
					                                                                                      Country/region [select]       Terms of use


                                                                                        All of dW

   Home         Products          Services & industry solutions     Support & downloads             My IBM




                                      developerWorks > Wireless >
developerWorks
More in this series:                  Wireless robotics: What, exactly, is
                                      "wireless robotics?"
Wireless robotics

In this article:
   What is wireless robotics?
                                      How to marry wireless technology with mobile robotics for a blissful
   Thoughts on mobile robotics        outcome
   A bot's-eye view of wireless
   technology                                                                                                                      Document options
                                      Level: Introductory
   Wireless robotics software                                                                                                          Print this page
   applications                       Erik Zoltan (erik@zoltan.org), Advanced Systems Engineer, EDS
   In conclusion
                                                                                                                                       E-mail this page
   Resources
                                      08 Feb 2006
   About the author
   Rate this page                                                                                                                  Using XML, but need to do
                                                                                                                                   more?
                                               This article, the first of a two-part series, takes a high-level look at wireless
                                               robotics. The article discusses what, exactly, wireless robotics is, looking at         Download DB2 Express-C
Related links                                  its two central components: mobile robotics and wireless technology. I will             9
   Wireless technical library
                                               also cover software applications of wireless robotics. Part 2 will examine
                                               some interesting applications of wireless robotics that you likely will hear
                                                                                                                                   Rate this page
                                               more about in the future.
                                                                                                                                       Help us improve this
                                                                                                                                       content

                                      What is wireless robotics?
                                      Wireless robotics is not a separate field of its own. Instead, it may be regarded as the timely confluence of two
                                      technologies: wireless technology and mobile robotics. Wireless technology is used to send information out using
                                      radio waves; this includes wireless computer networks. The proliferation of open wireless networks throughout offices
                                      and urban areas has big implications for robotics. Mobile robotics generally involves untethered robots that run on
                                      batteries and have an on-board computer.

                                      When you add wireless technology to mobile robotics, it opens the door to collaborative behavior between multiple
                                      mobile robots, and to the delegation of responsibility to an external source such as a central computer or a human
                                      user.

                                      I address what's meant by wireless robotics because some may still need clarification on the boundaries of wireless
                                      robotics and where it is headed.

                                      In 1898, Nikola Tesla filed a patent entitled "Method of an Apparatus for Controlling Mechanism of Moving Vessels or
                                      Vehicles" (see Resources). This patent covered his newly invented approach for using wireless radio signals to
                                      remotely control an external vehicle. He called these vehicles "teleautomatons." Tesla had invented a powerful radio
                                      transmitter, and by placing a receiver on the vehicle he was able to remotely produce the electrical signals needed to
                                      guide the vehicle. He even became concerned with wireless security issues -- the vehicle would only respond if a
                                      specific combination of frequencies was used to transmit the control instruction.


                                      Thoughts on mobile robotics
                                      Mobile robotics is a future technology in the same way that hand-held PDAs were a future technology two decades
                                      ago. Today, we have extremely powerful computers that, nevertheless, are very small and consume minimal battery
                                      power. Combine them with the latest small, lightweight batteries that can deliver power for quite a long time, and
                                      we're now primed to develop new applications for mobile robotics. However, the most critical ingredient, the ability for
                                      the robot to truly understand what it's doing, is still a long way off. As a result, there are several different approaches
                                      one can take to mobile robotics right now. (For the purposes of this discussion, let's simplify matters and assume that
                                      we're not using wireless technology.)

                                           1. Artificial intelligence. The fundamental problem is the lack of intelligence in a mobile robot. So working
                                              directly on the intelligence question makes sense. Vision processing is an example. If you can get a computer
                                              to process visual information and more accurately recognize and identify objects, then you can definitely
                                              improve the performance of a mobile robot.
                                               Note: The fact that some people might regard "artificial intelligence" as a term they'd rather not use only
                                               serves to underscore the point that it's a very difficult task to make robots more intelligent!

                                           2. Robotic hardware and software. Even though we don't know how to make the robot behave intelligently,
                                              there is still a lot of work that can be done on the hardware end. For example, we could try to create a
                                              walking robot with arms and hands that has delicate touch sensitivity and always remains in graceful balance.
                                              This is quite an engineering challenge, but a lot of progress has been made. By the time someone figures out
                                              how to make a robot intelligent, it will probably have an excellent body waiting for it.
                                           3. Creative application development. In addition to building robots that are effective for certain known
                                              applications, we can try to find unknown applications that our existing robots would be good at. Probably the
                                              development of the robotic vacuum cleaner (the most successful mobile robotics application so far) falls into
                                              this category. What other ideas are out there? Think about it and send me an email if you come up with an
       interesting idea. Maybe I'll print it here.
    4. Platform independence. The availability of software development environments that aren't tied to a specific
       hardware platform or even operating system would seem to be important for mobile robotics. Why? Because
       the mobile robotics platform of the future hasn't been invented yet. Someone will invent it, though, and there
       is no telling what processors, development packages, or operating systems will be available on such a
       platform. We need to be ready for anything.

This isn't intended to be an exhaustive list by any means. There are certainly a lot of other things one could work on
in mobile robotics, so think of these as a few broad areas that need attention right now.


A bot's-eye view of wireless technology
The preceding were some ideas that come to mind when you think about mobile robotics in isolation, without
considering wireless technology. But what if you do think about wireless technology in isolation, from the perspective
of mobile robotics? What if you consider those wireless applications because you are explicitly interested in mobile
robotics (as opposed to laptop computers, telephones, PDAs and so on) If you look at wireless technology, there are a
number of interesting avenues that come to mind.

    1. Off-the-shelf networking. 802.11x, Bluetooth, and other wireless technologies are implemented at a lower
       level and may have operating system-level support. Thus, they function seamlessly, allowing the system to
       behave as though the connections were wired. For example, I can write a TCP/IP application that would work
       the same way with a wireless network or an ethernet cable. My mobile robotics application might not make
       sense if the robot had to drag a long cable around, but the programming is all the same. This may sound
       obvious, but it is a major breakthrough in the field of mobile robotics. It also suggests a development
       approach: I can plug my robot into a wired network and write my code, and most of the time it will work just
       the same when I later switch to wireless. This suggests that the wireless robotics code you write today might
       continue to be compatible in the future, as new advances in wireless technology continue to emerge.
    2. Custom wireless technology. If I'm not satisfied with the limitations of off-the-shelf wireless technologies, I
       can always implement my own, as long as I obey relevant laws about broadcast frequencies and so on. Maybe
       I have a tiny embedded platform (such as a very small mobile robot running on a single microcontroller) and I
       don't want to bite off the overhead of a more-robust off-the-shelf technology, so I try to keep it simple. Or
       maybe I think that I can improve security by implementing an especially baroque form of encryption and not
       telling anyone how it works.
    3. Range enhancement. Numerous popular books and Web sites contain a lot of approaches to building a
       better antenna that will increase your range. These ideas work on the assumption that you are using a car or
       a laptop. But what if you are using a wireless robot? Doesn't this open up some possibilities that wouldn't be
       available with a laptop? For example, perhaps a wireless robot could use some kind of a directional antenna.
       There are various designs for directional antennas, but let's visualize it as a dish antenna similar to a satellite
       dish, only smaller. The robot could position itself at a location where the signal is strongest if it needs to do a
       large upload or download. It could align the dish in the position where the signal is strongest. While the robot
       is moving around, it could continually keep the dish aligned in the proper orientation to maximize signal
       strength. You might choose to have a second (non-directional) antenna to monitor overall signal strength. If
       there are two wireless networks available, and the robot moves closer to one of them, then it might want to
       switch to the other wireless network and re-align the directional antenna to point at the new signal.

Wireless robotics software applications
I've talked about the two fields in relative isolation from each other. Now we can combine wireless technology with
robotics, and there are some interesting possibilities that come up at the point where the two intersect. Let's begin by
thinking about software applications for wireless robotics.

Remote control
This isn't very surprising, because Tesla was already able to do this sort of thing at the turn of the twentieth century!
Many children's toys are surprisingly advanced remote control machines. However, there are still some significant
issues here that are worth mentioning.

In terms of simple control of movement, we can now be more advanced in terms of how we control a robot's
movement. I could wear a sensor glove, or a full body suit, and the robot might be able to mimic my movements with
great precision. But what about a robot whose body is nothing like mine? If we develop a very useful robot whose
movements are very unlike our own, then it will be a challenge to find a way to control it remotely. So the
development of intuitive user interfaces for complex remote control may actually still be an interesting area,
depending on what you are controlling.

Let's suppose that I am trying to control a spinning robot that goes around in circles at hundreds of RPMs. Obviously,
I'm not going to mimic its exact movements. Fortunately, in this case, I could use something like a joystick or a
mouse to tell it where to go. I could use something like a lever, a dial, or a foot pedal to control its rotational speed.

Imagine a lightweight window-climbing robot that looks like a ball covered in suction cups. Maybe there is an air hose
connected to each of the suction cups and it can inhale to stick to the window or exhale to let go. What kind of remote
control application would be appropriate to control this kind of a 'bot?

Imagine a scenario in which you want to control a larger number of robots working together. Examples of the same
everyday activities where humans typically cooperate include a soccer team, or maybe one robot holding the door
open for the rest to go through. It would be nice to be able to select individual robots and control them one at a time
if necessary. It would also be good to have some way to control a handful of robots at the same time, coordinating
them easily to work on a task together. There is a lot of work still to be done in this area.

One promising approach lies in the development of higher-level instructions. There are many very useful tasks --
ranging from simple to somewhat complex -- that are surprisingly easy to implement:

     q   keep going forward until you encounter an obstacle
     q   follow the white line until it ends
     q   keep climbing until you appear to have reached the top of the hill
     q   keep the camera focused on anything that moves

To the extent that you are able to add useful capabilities to a robot, remote control applications may become
progressively easier to use by combining many complex interactions into a single handy command.

Robot monitoring
Presumably, any good wireless robot is going to have some sort of sensor data that it can report back to an external
source. This could include streaming audio and video, positional coordinates, internal temperature, battery state,
touch sensor data, the position of any arms, legs, and other movable components, and so on.

If a human being is going to monitor this kind of information, then there is the challenge of presenting the information
to the user in a useful way. Does a temperature readout simply display a number, or does it look like a thermometer?
Does it flash or change color when the temperature is outside a certain range? Is streaming video turned off when no
one is looking, to conserve bandwidth? If there are a significant number of touch sensors, then how do you represent
this data in a way that is useful to the human user?

If the sensor information is being used to make decisions (either on board the robot or on a remote server), then
things can get rather complicated in a hurry. Reacting to sensor data is frequently a complicated affair. Even the most
simple things, like "back up and turn left when you encounter an obstacle", often turn out to be more difficult than we
assumed. (The sensors didn't get triggered by the obstacle; it encountered another obstacle while backing up; it fell
down a flight of stairs.)

So even when the robot is reacting independently to sensor data, it is helpful to keep a human user posted on its
activities and also to leave an audit trail. If something goes wrong, you will want to look back at the details as you try
to correct the problem.

Robot communication
Remote control assumes that the robots are talking to one or more human users; but here they are talking to each
other. The communication software itself wouldn't be much more complex than a chat room. The really tricky part is
figuring out what kind of useful information two robots can share.

Ultimately robot communication would have to arise out of a shared task. To take a simple example, let's suppose
that one robot is carrying a bunch of freshly charged batteries, and it needs to pass them around to the other robots.
The battery-carrying 'bot could ask the other robots to report their current positions. It would then choose one
(presumably the closest one with an old battery) and the two robots would either agree on a place to meet, or they'd
head in each other's direction, regularly transmitting their locations until they finally meet.

If a group of robots are working together to search for a hidden object, then they could all regularly transmit their
locations to a central server, which would keep track of a map of places that have already been searched. Gradually
the number of already-searched places would increase and the number of unsearched places would decrease until the
object is finally located. In this scenario, the central server could divide the search area into multiple zones and assign
one particular robot to look in each zone, or the individual robots could simply report their current positions and the
server would report back the nearest place that hasn't been searched yet.

Finally, imagine a soccer-playing team of robots similar to RoboCup (see Resources). To be successful, each robot
must be able to see the ball, its teammates, the opponents, the goal, and anything else in the field of play. Robots
need to be able to think strategically, follow the movements of the ball, and estimate its future location. This is some
pretty complex calculating going on. The robots should obviously communicate their identities and their positions to
each other regularly, as well as their immediate intentions for movement. There may be an external server that helps
to coordinate the robots into a "team plan." Essentially, though, the kinds of communication will vary based on the
quality of the information that is available, the physical movement capabilities of the robots, and the sophistication of
the strategic processing that is available.


In conclusion
That's an overview of the kinds of software that might be useful in wireless robotics. In Part 2 of this series, I will
spend more time on some interesting applications of wireless robotics. Maybe I can discover the "killer app" that will
set wireless mobile robotics on fire (not literally, of course).


                                                                                                              Back to top


Resources
Learn
     q   See all the articles in this Wireless robotics series.

     q   Check out RoboCup if you're interested to learn about how soccer-playing robots work and how they
         communicate.

     q   Learn about IBM's contribution to robotics.

     q   Read Nikola Tesla's patent entitled Method of an Apparatus for Controlling Mechanism of Moving Vessels or
         Vehicles.

     q   Read this previously published article "Introduction to robotics technology" (developerWorks, September
         2001).

     q   Stay current with developerWorks technical events and Webcasts.
                       Get products and technologies
                            q   Download a free trial version of WebSphere® Application Server Version 6.0.

                            q   Build your next development project with IBM trial software, available for download directly from
                                developerWorks.


                       Discuss
                            q   Participate in developerWorks blogs and get involved in the developerWorks community.


                                                                                                                                    Back to top


                       About the author
                                      Erik Zoltán got his start in AI in the late 1980s, with the invention of a proprietary neurosemantic
                                      network technology that is still under development. His current work focuses most strongly on
                                      computer perception. He gradually became involved in robotics when the field kept coming up as an
                                      important application of his work. He has taught programming to hundreds of professional developers
                                      and has worked on CBT, databases, and communication systems in industries such as banking,
                                      education, healthcare and government.




                                                                                                                                    Back to top


                       Rate this page
                        Please take a moment to complete this form to help us better serve you.

                        Did the information help
                        you to achieve your goal?       Yes             No           Don't know



                        Please provide us with
                        comments to help
                        improve this page:




                        How useful is the
                        information?                    1           2           3            4           5
                        (1 = Not at all,
                        5 = Extremely useful)




                                                                                                                                    Back to top


About IBM   Privacy   Contact
                                                                                   Country/region [select]     Terms of use


                                                                                    All of dW

   Home         Products        Services & industry solutions     Support & downloads           My IBM




                                    developerWorks > Wireless >
developerWorks
More in this series:                Wireless robotics: What, exactly, is
                                    "wireless robotics"?
Wireless robotics

In this article:
   Applying wireless robotics
                                                                                                                               Document options
   Telepresence                     Level: Introductory
   Telepresence robotic services                                                                                                    Print this page
   Autonomous robots                Erik Zoltan (erik@zoltan.org), Advanced Systems Engineer, EDS
   Wireless Network Applications                                                                                                    E-mail this page
   Convergence
                                    29 Mar 2006
   Resources
   About the author                                                                                                            Using XML, but need to do
   Rate this page                                                                                                              more?
                                             Examine some interesting applications of wireless robotics that you may be
                                             hearing more about in the future. This article covers several areas that               Download DB2 Express-C
                                             appear to be major applications of wireless mobile robotics.                           9
Related links
   Wireless technical library
                                                                                                                               Rate this page
                                    Applying wireless robotics                                                                      Help us improve this
                                                                                                                                    content
                                    This second part of a two-part series outlines wireless robotics at a high level. The
                                    first part gave a definition of the field and talked about some software-level
                                    applications. It described how advances in wireless technology, the availability of very small high-performance
                                    computer systems, and the low-cost availability of robotic components are making this a good time to start
                                    developing wireless robotic applications. This article takes more of a "big-picture" view of the field and talks about
                                    macro applications that suggest themselves.

                                    Remote control
                                    A remote-controlled robot naturally begs the question: Why would I use a remote-controlled robot instead of just
                                    doing the same tasks by hand myself? For example, a remote-controlled robot to comb a child's hair would probably
                                    be a lot harder and more expensive than just walking over and combing their hair by hand. Perhaps this is an absurd
                                    example, but it illustrates an important point: There is a difference between a can and a should. I can make a large
                                    investment of time and capital for certain robotics applications, but which ones should I choose and which ones would
                                    be better to ignore?

                                    I can think of a number of reasons why I'd want to use a remote-controlled robotic device instead of doing something
                                    it by hand.

                                    Danger. The robot can go into a hostile or extreme environment (such as an environment where radiation or
                                    contaminants are present, extreme temperatures, or an unstable building where an earthquake has occurred) where
                                    it would be unsafe to send a human. Robots are already being used in combat environments or search-and-rescue
                                    missions.

                                    Scale. The robot may need to do things that are much too small and delicate, or too large and heavy, for a person to
                                    do by hand.

                                    On a large scale, large vehicles such as cranes or fork lifts are already being used to manipulate heavy objects, and
                                    perhaps there is room for robots in this domain as well. If a giant crane is currently being used to elevate materials up
                                    to the top floor of a building where humans can work with them, then imagine how much more useful a giant human-
                                    operated robot would be for the same activity. (But, perhaps some people would find this to be an alarming prospect.)

                                    On a small scale, a tiny little robot can take large scale movements of the human hands and reproduce them on a
                                    smaller scale, allowing a person to use natural hand movements to do something very small and delicate. We already
                                    have some applications that fall into this category, and I expect a lot more. As miniature hardware components
                                    continue to improve, there is definitely going to be more progress in this area. Think of the surgical applications, for
                                    instance. Currently, it's possible to swallow a capsule that wirelessly transmits information, such as the core body
                                    temperature, to an external receiver. Imagine swallowing a small robotic capsule remotely controlled by a surgeon. As
                                    it enters your intestine, the surgeon might cause it to unfold itself and remove a polyp or repair a hernia.

                                    Location. Sometimes a person can't be in a specific location for practical reasons. Maybe it's too far away, or they
                                    simply can't be there around the clock. A robot could remain in that location as an onsite agent. This can be done
                                    already at a purely virtual level with telephones, e-mail, or videoconferencing. However, a robot would add a physical
                                    dimension to this notion of "remote presence." This brings me to my next topic.


                                    Telepresence
                                    Telepresence might just be the first "killer app" for wireless robotics.

                                    Telepresence can be thought of as transmitting your physical presence to another remote location. (see Resources for
more information). The idea is that even when you are physically in one place, you are experiencing sights and
sounds, and controlling movements in a natural way, somewhere else. In terms of wireless robotics, you would have
an untethered robot at a remote location, equipped with video and audio sensors, and a speaker. So the user could
see what the robot is seeing and hear what the robot is hearing. You could speak into a microphone and your voice
would be transmitted by the robot. It's important that the remote controls should be highly responsive and very
intuitive to use. (Otherwise, it becomes harder to use the word "telepresence" with a straight face.)

A wheeled robot would be okay for telepresence, particularly at first. However, a bipedal robot (if it can be controlled
using natural body movements) would be the most ideal.

A telepresence robot called Rokviss is currently being used on the international space station. It can be remotely
controlled from the ground, but only for a few minutes at a time, while it is over the receiving station in Germany.
However, it's a simple research prototype that doesn't meet the full-blown definition of telepresence that I'm
presenting here.

Telepresence robots have a big advantage over many other applications: the human is fully in control, so there's
really no need to solve the problem of intelligent autonomous behavior. Difficult issues like balance, vision, speech
recognition and understanding the world around the robot simply aren't a problem in telepresence applications. This is
the primary reason why it's one good choice for an early "killer app" of wireless robotics.

There are several challenges in telepresence robotics.

     q   You need a user interface that lets the user see, hear, and communicate in a realistic way and control the
         remote robot intuitively without having to think about what they are doing.
     q   You need a robot that mirrors the user interface exactly, and that responds as intended to the control signals
         that have been received. So, as you enhance the hardware robot, you may have to update the user interface
         to keep it in sync, and vice versa.
     q   The robot also must give appropriate feedback when problems are encountered.
     q   You need software to integrate the user interface with the robot, that is, to transmit and to transmit sensory
         information from the robot back to the user.
     q   Finally, you want to stay connected. So you must have a good wireless link between the robot and the user.
         When the communications link is inevitably broken (and this is virtually guaranteed to happen), you need
         some kind of a fallback strategy.

Here are some thoughts on applications of telepresence robotics.

Stand-in robot. This is a robot that stays somewhere you can't be. For example, it can be remotely controlled and it
can be in your office. It can type on your computer for you or answer the phone with your voice. Even though you're
not present and there are certain things that must be done physically, you have a remote-controlled agent that is
capable of doing them for you. If your office has wifi the agent can walk all over the office, making copies, sending
faxes, and so on.

Imagine you're a one-person company, but you're away on a business trip or on vacation. You can't be there to open
and read the mail, do some unexpected filing or photocopying, read a fax that just arrived, or open that file cabinet
and look at a copy of a document that only exists in paper form. As technology advances there are fewer things that
you need to do physically on location, but they still exist and they are still important. Did you ever have a server in
your office go down in the evening when no one was there to reboot it?

Or, you can look at it from the opposite perspective: Imagine you're in the office and your telepresence robot is at
home.

People are investing money to get their homes and household appliances all connected to a computer. For example,
perhaps I want to buy a coffee pot that can be programmed remotely. Maybe I can sit at the office with my laptop
and tell the coffee pot to start in 45 minutes so there will be a fresh, hot cup of coffee waiting for me when I get
home. The problem with this kind of thing is that each appliance is much more expensive than an equivalent product
that I probably already own. How am I going to convince my wife that we need a new coffee pot, a new thermostat
that I can control from my laptop as I'm sitting on the couch, and a Web-enabled remote car starter for those frigid
New England mornings? What about having a telepresence robot instead? Why not centralize my financial investment
in the form of a wireless robot that can perform a variety of household tasks? In addition to having it program the
coffee pot to go off at a certain time, I might be able to somehow figure out how to make it grind the beans, change
the filter and pour the water.

There are other things I could do with this kind of telepresence robot. It could feed the dogs and let them in and out
of the house without granting access to strangers. It could adjust the thermostat, turn on the bread machine, empty
the litter box, or leave a post-it note on my wife's car keys reminding her that the car is low on gas.

Why make a separate investment in a variety of task-specific household devices when you can get a single
telepresence robot that does it all?


Telepresence robotic services
Telepresence robotic services is an extension of telepresence robotics. After you have a telepresence robot in your
home or office, why not hire someone else to control it for you?

Imagine that there is a robot in your home and it can be controlled remotely over your wifi network by a personal
service provider who knows the special security keys required to control it. This robot could take care of your house
while you are away on vacation (feed the pets, water the plants, keep the home secure). The same things you'd have
a house sitter do for you. Such a service might be a lot less expensive than having a human being actually come to
your home, especially if you have an over-protective dog. The services that this kind of robot could provide would
depend entirely on the physical capabilities of the robot.

Wheeled robots:
     q   They move over level terrain and are easily controlled remotely.
     q   May not be so good on stairs.
     q   Getting stuck on uneven terrain may be an issue.

Cameras:

     q   Can provide visual feedback, allowing a remote service provider to see exactly what they are doing.
     q   Useful for security and monitoring.
     q   A robot with a camera can move around through an area and see every corner, so it's not subject to the
         limitations of a stationary camera.

Positioning:
You could use GPS or some other positioning technology if it's not clear where the robot is located. This would be
especially helpful if the robot is not inside a home or office -- maybe it's in the woods or going down a road.

Legged bots:

     q   May be better at stair climbing, depending on the design, but are generally harder to coordinate than wheeled
         robots.
     q   Tripping or falling over may be an issue.

Arms/hands:

     q   Gives the robot the ability to grasp and manipulate objects. This would be useful for many remote applications.
     q   Manual sensitivity and dexterity are critical, as is the development of a good user interface for remotely
         controlling the environment.

Imagine a remote-controlled robot that can change recessed or hard to reach lights.


Autonomous robots
Here are some ideas for applications of autonomous mobile wireless robots.

     q   "The Scoop," a small autonomous robot that runs around trying to gently scoop up any small objects it comes
         across and places all the objects in a box. I have emphasized the words gently and small because some
         objects are more delicate than others and because a sleeping pet might not take kindly to being handled in
         this way.
         It could have a large scoop in front, and it could patrol the floor space, backing up when it bumps into an
         object that doesn't get scooped up. When something lands in its scoop (such as a scrap of paper, a pen that
         has fallen to the floor, or even an old sock), it can simply place that object in a special box for a human being
         to organize later.

         I once developed a simple prototype of this idea in my home, and my two small boys thought it was very
         funny to sit in the path of the robot or to place other obstructions in its way to see how it would react. I
         encouraged them to do this, and often found myself tweaking the robot or rewriting the driver code in
         response to humorous jokes they had managed to play on it.

     q   A robot that is capable of locating a wall outlet and plugging itself in when it needs to recharge. This is not a
         major application by itself, but if you can create such a robotic platform then it can serve as a base system for
         others to use in developing applications.
     q   Vacuums and lawnmowers already exist, but we can generalize beyond these specific applications. They are
         both instances of the general class of activities that I'll call "walk around and do menial chores." A number of
         other extensions of this approach seem to naturally suggest themselves.
             1. Water the plants. You might have special planters or maybe wireless transponders to facilitate doing
                this autonomously.
             2. Control the household temperature by opening and closing windows and curtains. One would certainly
                need to install the right kind of window and curtain hardware to enable this application. Even so, you
                only need one motor and control system (on the robot) instead of having one for each window and
                curtain. Instead of having a temperature sensor in each room, the robot could carry a single
                temperature sensor with it. Plus, you'd need one for outside. On the downside, the robot needs to
                navigate to each window and curtain, which wouldn't be necessary in a traditional home automation
                application. You'd want to simplify this process with a radio signal or something else.
             3. Monitor environmental conditions. This is an extension of the previous idea. Stationary sensors in the
                home can monitor temperature, detect smoke or carbon monoxide, radiation, and so on.
                Alternatively, a robot can have its own sensors or carry portable sensors to any location it can reach.
                To understand why this might be useful, imagine you have a source of a minor amount of carbon
                monoxide in your home. Even though the overall levels are within acceptable ranges, a mobile robot
                might detect high levels in an isolated area that might not be picked up by a stationary sensor.
             4. A mobile robot could serve as a remote car starter. Of course, a human user could pilot it remotely,
                but this would be most desirable if it were an autonomous application. Other applications fall under
                the same heading such as adjusting the thermostat shortly before you arrive home.
             5. Taking care of the pets. Imagine a robot that can make sure the food and water dishes are filled at a
                certain time each day, or even empty the litter box regularly.
             6. Taking out the trash. You might need a special trash can, but some people would consider this a small
                price to pay to avoid this stinky and menial chore.
             7. Preparing food. Hey, don't laugh at this one. Perhaps in a decade or so, there will be a section of your
                grocery store devoted to specially-packaged foods that can be autonomously prepared by your mobile
                robot. It's just a new kind of convenience food, and it's a natural extension of recent history. First
                there was frozen pizza for the oven, then there was toaster oven pizza, and now there is microwave
                oven pizza. So why not robopizza? This would be pizza (and other types of food) in special packaging
                that includes a computer chip that can tolerate freezer temperatures. Your mobile robot will simply
                open and close the freezer, locate and retrieve the pizza from a special compartment in the front,
                remove it from the custom-designed packaging, and pop it in the oven. It's really not that big a leap
                from a bread machine.
             8. Delivery. Why have a person bring samples to the lab for analysis when a robot can do it more
                patiently and without getting sidetracked? There are a lot of delivery applications that could be
                performed by a robot. Of course, there are also a lot of delivery tasks that still have to be done by a
                human. A robot can't deliver a pizza and obtain payment from a customer yet because there is too
                much human-level interaction involved and too much navigational uncertainty. But anything where
                these conditions don't exist might be a good candidate for robotic delivery. If it is easy to navigate
                from point A to point B, and there is no doubt about how to get through a locked door or exactly
                where you are going, then most likely a robot can be programmed to get there. If it is easy to make
                the delivery (for example by leaving the cargo in a specific place that is guaranteed to be available or
                having a person who is guaranteed to be there and will know how to retrieve the cargo from the
                robot), then an application might be a candidate for robotic delivery.
             9. Development/research platforms. When there is an exciting technology and nobody knows exactly
                what to do with it, there is room to create a base system that doesn't actually do very much, but
                already contains some of the ingredients that a useful system would need in the future. Then you can
                sell this base system (even though it's basically a "useless shell") to somebody else who is also
                working in the field. This makes a great deal of sense. Each developer or researcher is spared the
                agony of recreating the wheel and resolving many of the core problems. The prospective developer
                can focus directly on solving a single more-advanced (and more-interesting) aspect of the domain.

Most of the above applications should be autonomous for maximum benefit. Generally, they aren't good candidates
for remote control unless you can't to be there to do them personally or if you have physical limitations that makes
them a hardship.


Wireless Network Applications
If you are using a wireless robot over your normal wireless network, then there are a number of applications that are
related to the wireless network itself. For example, the robot could exhaustively patrol every inch of territory inside
your home or building and create a map of what networks are accessible everywhere, and what the signal strength is.
(Actually, you don't even need to have your own wireless network for this to be a useful application.)

The robot can even move around outside the building, say in sidewalks and parking lots, or right outside another
company's office, to see if your network is unintentionally available outside the physical boundaries that you thought
it had. Maybe you have a client who doesn't mind that you share their data over a wireless network because you have
assured them that the network cannot be accessed outside of the building. With this robotics application, you can
have a great deal of confidence that it's really true.

Or, maybe you have a client that won't permit you to share their data over a wireless network for security reasons.
Imagine that you've assured them that there are no wireless networks in your buildings, and that the wireless
network card has been disabled on all employee laptops. Are you sure that's really true? A robot that walks around
the office routinely searching for wireless network traffic could help you to avoid an embarrassing situation. As soon
as it detects a wireless signal, you'll know that someone forgot to turn off their wireless adapter and you can remind
them right away before any problems can arise.

Another idea is a wireless network extension. What if you are outside and you need to stay connected to a network? A
series of mobile robots could follow behind you and remain connected to the base network. They could stay behind at
various intervals to create a "daisy chain" router network.

Why not just carry a series of wireless routers and leave them behind as you go? An autonomous mobile robot could
move around to various areas in an effort to get the best possible signal. In this way you could get a better result
than if you tried to manually position a series of antennas. Also, autonomous robots could sense changes in your
position and move dynamically to make sure you stay in range. And if you simply forget about them, they could
independently return back to the base station when their batteries start to get too low.


Convergence
Many people are talking about the integration of multiple devices into a single device; for example, your television,
your computer, your handheld computer, your cell phone, your ISP, your telephone, and your entertainment system
(CD/VCR/DVD/stereo). All of these separate devices are now converging into a single device.

But why not add robotics to this equation? Why should your television stay in one place? What if your "surround
sound" entertainment system could follow you around the house? What if that elusive remote control could rise up
from between the couch cushions to answer your summons? What if pushing the "call button" on your cordless
telephone system could cause the handsets to return back to the base station instead of merely beeping at you?

What if your appointment book and task list could walk over and tap you on the shoulder to remind you of an
important obligation?

As the underlying technologies continue to come down in price, many of these applications will start to look more
practical.

Conclusion
I hope that this discussion of wireless robotic applications has gotten you as excited as I am about the future of
wireless mobile robotics. The proliferation of laptop computers, the internet, cell phones, and wireless technology has
certainly revolutionized everyday life in recent decades. Mobile robotics has a potential to extend this revolution into
the physical arena. Even though robotics is still in its early stages, and many problems remain to be solved, there are
still some interesting applications just over the horizon that could easily enter your everyday life in the next decade or
so.


                                                                                                              Back to top


Resources
                       Learn
                            q   Transparent Telepresence Research Group: Telepresence can be thought of as transmitting your physical
                                presence to another remote location.

                            q   "What, exactly, is "wireless robotics?" (developerWorks, February 2006): Read the first article in this two-part
                                series.

                            q   Wireless robotics: Read all the articles in this series.

                            q   Telepresence: Here is a link to a popular article about telepresence.

                            q   IBM robotics: Learn about the IBM contribution to robotic technology.

                            q   "Introduction to robotics technology" (developerWorks, September 2001): Read this previously published
                                article to find out more about robotic technology.

                            q   developerWorks technical events and webcasts: Stay current with the latest technologies.

                       Get products and technologies
                            q   IBM trial software: Build your next development project with trial software, available for download directly
                                from developerWorks.


                       Discuss
                            q   developerWorks blogs: Get involved in the developerWorks community.


                                                                                                                                    Back to top


                       About the author
                                       Erik Zoltán got his start in AI in the late 1980s, with the invention of a proprietary neurosemantic
                                       network technology that is still under development. His current work focuses most strongly on
                                       computer perception. He gradually became involved in robotics when the field kept coming up as an
                                       important application of his work. He has taught programming to hundreds of professional developers
                                       and has worked on CBT, databases, and communication systems in industries such as banking,
                                       education, healthcare and government.




                                                                                                                                    Back to top


                       Rate this page
                        Please take a moment to complete this form to help us better serve you.

                        Did the information help
                        you to achieve your goal?         Yes              No              Don't know



                        Please provide us with
                        comments to help
                        improve this page:




                        How useful is the
                        information?                      1            2            3            4        5
                        (1 = Not at all,
                        5 = Extremely useful)




                                                                                                                                    Back to top


About IBM   Privacy   Contact

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:1923
posted:4/1/2008
language:English
pages:9