Добавил:
ivanov666
Опубликованный материал нарушает ваши авторские права? Сообщите нам.
Вуз:
Предмет:
Файл:Readings in Robotics Engineering. Учебное пособие
.pdf
1.3. What makes a robot a robot?
11
• Is a dishwasher a robot? Why or why not?
• Is a cellphone a robot? Why or why not?
https://www.fi.edu/sites/default/files/EducatorGuide_RobotRevolution_Ex
hibit_edguide-robot-revolution.pdf
Choose a suitable robot for each scenario.
Scenario 1
The local hospital wants to make children feel more comfortable after
they come out of surgery. Unfortunately, they cannot bring live animals like a
dog or a cat to help relax the young patients, but they can bring in robots and
different types of artificial intelligence. What type of robot would you offer to
the hospital?
• How does the robot work?
• How does your robot benefit the young patients in the hospital?
• Why is this robot the best choice compared to the other robot options?
Scenario 2
A new electric car company has just been funded and they are starting to
put together their facility where the cars will be produced. The company has
hired several employees but they are quickly learning that some of the car parts
are too heavy for one person to lift. What type of robot would you offer to assist the electric car company?
• How does the robot work?
• How does your robot benefit the young patients in the hospital?
• Why is this robot the best choice compared to the other robot options?
Scenario 3
Scientists are interested in researching rock formations at the Grand
Canyon. As they are exploring they discover a cave and have come to a point
where humans can no longer fit through the opening. They are curious about
the depth of the cave, and if there are any living plant life or insects present.
What type of robot would you offer to the scientists to enhance their research?
• How does the robot work?
• How does your robot benefit the scientist researching the Grand Can-
yon?
• Why is this robot the best choice compared to the other robot options?
Part I. Sensing the World
12
Scenario 4
NASA is interested in learning more about Jupiter’s moon Europa.
Scientists have discovered that under the icy surface of Europa there is an
ocean they would like to research and study. A mission this far into deep
space has never been attempted before and NASA is starting to plan for this
future event now. What type of robot would you offer NASA to send to the
Jupiter’s moon Europa?
• How does the robot work?
• How does your robot benefit the young patients in the hospital?
• Why is this robot the best choice compared to the other robot options?
Scenario 5
A large Corporation sells and ships items all over the world. They sell
items as small as a ring to as big as a car. They can get many items to their
customers in as quickly as 24 hours, but would like have packages delivered
in less than 12 hours. These packages sometimes need to get across oceans or
mountains.
Scenario 6
A small country is constantly threatened by earthquakes. Recently a
7.8 magnitude earthquake hit the country, killing hundreds of people and de-
stroying many homes and businesses. The country’s leaders would like a robot
or robots that can search through rubble to find survivors, help with the clean
up and rebuild structures to get the country back on its feet.
Scenario 7
Farms all over the world are attacked by bugs, small mammals and birds
that feed on the crops. Many farmers don’t want to kill these animals, but
would prefer if they found their food somewhere else. They would like a robot
or robots to help; these robots can be any size, but cannot interfere with the
farmers when they are harvesting their crops.
Scenario 8
An elderly woman with no children is starting to have trouble getting
around her house. She is starting to forget to take her medicine and needs help
preparing her meals. She wants to be able to stay living at home, but is going to
1.3. What makes a robot a robot?
13
need help. The robot needs to be able to easily move around her house, help
with everyday chores and give her reminders.
SWARM ROBOTS Inspired by
flocks of birds and swarms of insects in
nature, swarm robots work together in
groups to operate. Each robot performs its
own task, but they all follow one leader in
the pack. These robots are currently in the
research phase. Potential future uses include search and rescue, mining and
even miniaturization for medicine.
RHex has the ability to travel
through rocks, sand and other climates.
This robot is used to study areas that humans are not able to reach or that are unsafe
for travel. Attaching objects like climate
sensors allow RHex to collect data for hu-
mans to study later.
The M-1iA Delta Robot works in assembly lines to increase the production time
of a product. Humans use hand-eye coordination to complete tasks on an assembly line.
Robots are very precise and their vision,
powerful motors (actuators) and lightweight
arms can work more efficiently than humans. Today it is more likely for a
worker to be trained how to program and function a robot to work on an assembly line rather than completing the task themselves.
PARO is a therapeutic robotic seal. It
is designed to look approachable and to have
a calming effect on people who interact with
it. PARO works with the help of interactive
sensors that allow it to respond when being
touched. PARO will respond to petting with
reactions such as moving its tail, “purring” and closing its eyes.
EMYS stands for “Emotive headY System.” It was designad to research
how humans react to robots that show emotions. EMYS can show emotions like
Part I. Sensing the World
14
happiness, sadness, surprise and anger by
moving its three disks and two eyes.
EMYS can respond to “seeing” a person’s face and also responds to touch.
BAXTER is a robot that has the
ability to “learn.” A worker moves Baxter’s arms in a desired motion; Baxter
memorizes the motions and repeats them
over and over. Baxter is often used in
industrial environments and is considered unique because it does not need a
software engineer to program it.
The Da Vinci Surgical System
is a robot that assists in making major
surgeries minimally invasive. Its robotic arms carefully perform the surgery
on the body as the surgeon orchestrates
every movement, incision and suture
from the Da Vinci console. This con-
sole produces a three-dimensional, high-resolution image for the surgeon to
observe and manipulate while performing the surgery.
Unmanned aerial vehicle.
This robotic vehicle flies without a
pilot. It is controlled by people on the
ground or pre-programmed with missions before take-off. A drone can
sense and respond to its surround-
ings, such as wind, bad weather, or obstacles in its path.
Drones are used for many jobs, including farming, photography, military
surveillance, search and rescue, and supply delivery.
https://www.fi.edu/sites/default/files/EducatorGuide_RobotRevolution_E
xhibit_edguide-robot-revolution.pdf
1.4. The principles of sensing
15
INTERPRETING THE ENVIRONMENT
How do we make sense of the world around us? Whilst this problem
may sound easy to us, it always proves an issue for robots. We look at some of
the ways that robots can build a picture of their environment.
1.4. The principles of sensing
Sensing is one of the key pillars of robotics, leading to perception of the
world, then cognition and action. Robots use sensors to acquire data. But how
do they do this, and how do they process the data? This step introduces some of
these important principles of sensing.
Why do we need sensors?
Let’s start looking at sensing by asking two simple questions: Where are
you? And how do you know where you are?
These questions are relatively simple to answer for a person.
We are able to understand my environment and make decisions based on
very abstract concepts that we observe. We can observe what’s happening
around us, and use our senses to orientate ourselves in the environment based
on that information.
For a robot, the process is very similar. The first step is to acquire raw
data from sensors. The second step is to organise and understand that data,
using perception. Whilst this problem sounds easy to us, it is often challenging to robots.
Computers are the robot brain, and they’re digital
The first step in sensing involves transforming the world into something
that the brain of the robot, the computer, will understand.
Computers are digital, meaning that at the very lowest level they only
use 0s and 1s to represent data. All operations simply involve shifting these two
values around. Any representation of the world, at the very simplest level, must
be built from these values for a computer to make sense of it.
Therefore, the act of sensing can be thought of as transforming the real
world into a discrete sequence of values, or samples, that represent a particular
quantity (e.g. images, audio, velocity, temperature etc.).
Part I. Sensing the World
16
In the process of sensing, we sample the environment and provide a
snapshot of the world at that moment in time. As we cannot sample everything
about the world, this is not a complete model. It only represents the local environment, and probably isn’t even a complete representation of that.
Sensing for a stranded robot
Imagine a robot stranded somewhere in the world. What sensing process
can the robot use to understand where it is?
GPS. In the outdoor world, the Global Positioning System (GPS) is the
most common and easy to use technique to estimate position. By measuring the
time delay to a collection of satellites which are fixed in orbit above locations
on the globe, and using triangulation, a position can be determined to an accuracy of a few metres.
GPS is sufficient to give an approximate location (to an accuracy of
about 3.5 metres). This is useful but could be problematic e.g. if we wished it to
cross even a moderately wide footbridge. It also does not provide any information about the local environment so we could not avoid a person walking
across the bridge.
Therefore we need to get more detailed information about the surroundings. Typically this is through using devices like RADAR (Radio Detection and
Ranging) or even better, LIDAR (Light Detection and Ranging).
RADAR. Radar transmits a radio frequency beam which is reflected
when it comes into contact with an object. By measuring the reflections, objects
can be detected.
This system doesn’t scale well to small objects so doesn’t solve the nav-
igation problem for our robot.
LIDAR. Lidar is an object detection method similar to radar, which uses
light instead of radio signals to detect objects. Lidar measures reflected light
emitted by a laser, which has a much smaller wavelength than radio signals,
and so lidar can detect smaller objects than radar.
So, to solve the problem of our stranded robot, we can use a combination of sensing: GPS for approximate localisation in the world, radar for detecting large objects nearby like a building, and lidar to detect smaller objects, like
people. These sensors can be combined using sensor fusion algorithms to build
a complete picture of the world.
1.5. How do drones sense the world? Video
17
Discussion
What factors do you think could impact the reliability of sensors in the
real world?
1.5. How do drones sense the world? Video
In this video, we join Dr Jonathan Aitken and Dr Owen McAree in their
Sheffield Airport hangar to find out how they have designed an Unmanned
Aerial Vehicle (UAV) - more commonly known as a drone - to be able to sense
its environment.
They’ll explain some of the different kinds of sensors that have been fit-
ted to the drone so that it can build a picture of its surroundings and explore
some of the ways this could be useful in the real world.
Discussion
Can you think of any other useful applications for this type of drone?
Transcript
0:05 So, typically, you’ll have a sensor setup, where you’ll have some
way of sensing the world. That’s then passed into an autopilot. That autopilot
takes that information and builds the best picture of the world that it can. It will
be using information, for example, from a magnetometer to give a heading, or
from a GPS unit to give you a global position, or from accelerometers to give
you acceleration in various different directions. It takes all that information and
puts it together to give a rough picture of how the aircraft is moving at that
time. That means you can then take decisions based on how you want to control
the aircraft based on that information. This is our massive octocopter aircraft.
0:42 All the flying weight, at the moment, of this is around about
10 kilogrammes, but it will actually fly up to about 15 kilos. So it can carry a
really heavy payload. It started life as a standard octocopter. It’s got an autopilot
system on here that lets you fly it around using one of these controllers, just fly-
ing it around manually, like most people would do. But we’ve actually decided
that it’s quite difficult to do that. So what we’ve done is we’ve added a laser
scanner onto the front of it, and then actually attached some additional computers
onto the bottom. And this lets us do some really advanced control systems with it.
Part I. Sensing the World
18
1:21 Meaning that instead of just flying it around manually, so instead of
just having to fly up near a wall with manual control - which, when you’re
standing quite far away, can be very difficult and challenging to do - the laser
actually constantly tells the autopilot how close it is to the wall, and lets the
autopilot automatically choose to stay a fixed distance from the wall. So it
makes the pilot workload really, really low in comparison to what it would otherwise be, and allows you to use systems like we have on the front here - which
is a little camera system - to actually then go up and inspect things really closely without risking the aircraft.
1:55 And this is particularly useful if you’re flying outdoors in windy
conditions and all sorts of things like that. There’s already a lot of people who
are using these drones for doing things like structural inspection. So, flying it
really close into buildings and getting really good zoomed in views, because
it’s much safer and much easier and more convenient to do it with a drone than,
for example, put up a lot of scaffolding or have people abseil down a bridge or
something like that.
2:18 But the people flying those drones can find it quite hard work,
because in windy conditions and in that sort of environment, it can be really
difficult to fly these things close in to get the sort of images that the surveyor
then needs to actually make decisions about the structural integrity of the
structure. So, by adding these additional sensors and additional algorithms
and routines, it allows us to use these systems much more easily, and also in
much more adverse weather conditions, where a human pilot simply wouldn’t
be good enough. Because a human pilot has to watch the aircraft and wait for
it to move from a gust of wind or whatever, and then see that and then respond and then react.
2:59 Whereas, by using this system, before the pilot’s even noticed
something’s gone wrong, the aircraft has already corrected and stopped the
aircraft getting too close to the structure.
Reading: ROBOT BODY
In robotics, an end effector is the part of the robot that interacts with the
environment. The three most common types of end effector are mechanical,
magnetic and vacuum. End effectors are designed very differently depending
on the task they are intended to perform. Industrial grippers come in many dif-
1.5. How do drones sense the world? Video
19
ferent forms. Some physically grasp the object to be manipulated, like the
“hand” the students will build, while others puncture objects with needles, or
suck them up with a vacuum, or grip the object with glue. Some use electromagnets to pick up and drop magnetic objects. Effectors can also be tools like
drills, screwdrivers or welding torches.
Vacuum end effectors like the one above
have a granular material like coffee
grounds or sand inside and use a vacuum
to wrap around an object and then pick it
up. They can pick up delicate objects
like an egg without breaking them. However, they cannot pick up really large
objects or very tiny objects and are limited in how much they can move the object around once it is picked up.
Mechanical end effectors, like this threefinger gripper, grip objects similar to how
a person would. This allows for more
movement of the object once it is picked
up. However, they are not able to pick up
objects of every shape and will likely
crush very delicate objects like an egg.
Magnetic end effectors use a powerful
magnet or magnets to pick up or move
objects. They can pick up an object of
almost any size or shape, but the object
they want to pick up needs to be magnetic for it to work
Wheels allow a robot to move around
quickly and in all directions. They are
easier to make and program, but are limited to use on mostly flat surfaces.
Part I. Sensing the World
20
Two, four, six or more legs give a robot
movement similar to humans or animals.
Legs are hard to make and complicated
to program. Currently most robots with
legs move slowly and fall down easily.
Some can climb stairs and some can go
over obstacles, but most only work on
flat surfaces.
Propellers allow a robot to zip through
the water or fly through the air. They can
be damaged by running into an obstacle.
https://www.fi.edu/sites/default/files/EducatorGuide_RobotRevolution_Ex
hibit_edguide-robot-revolution.pdf
Reading: ROBOT SENSES
Identify human senses and how we use them for different tasks. Define
the difference between a sense and a sensor. Make lists of senses and sensors in
your notebooks. Read and check your suppositions.
We humans are experts at using our senses. A sense is a way in which
we perceive our surroundings. Humans have a sense of sight, sound, touch,
smell, and taste. A sensor is a device that detects something; for example, your
nose is the sensor you use in your sense of smell. Robots are machines that can
sense, plan, and act. To be a robot, a machine needs sensors. Some of these
sensors mimic or enhance human abilities, and some sensors do things that humans can’t perceive. Robot sensors allow them to reach beyond human capabilities in order to achieve new tasks, like exploring the surface of Mars or delivering packages across the globe in record time. When designing and building a
robot, engineers often use human and animal models to decide which sensors
that robot might need. A model is a representation of something that can be
used or applied to better understand a concept. For example, engineers designing a sounddetecting robot might use bats as an animal model for echolocation.
Соседние файлы в предмете [НЕСОРТИРОВАННОЕ]