Robot Helper Design Challenge
May 1, 2003
The design challenge WebCast.
We received lots and lots of designs and our NASA experts looked at them
We received designs from all over the United States, from Canada, from
Spain, from Australia, south Africa, Singapore, west INDIES.
They looked at all your designs.
Our experts are here today.
If you have any questions go to the chat room, type in the questions and
we'll answer them as many as we can during the WebCast.
So, our experts are going to look at your designs and comment on them
in each of the five areas.
They're going to look at propulsion, how does that robot move around the
And they're going to look at navigation.
How does the robot avoid obstacles?
How does the robot measure temperatures?
Where are the senses to measure air pressure and temperature?
How does it communicate with the astronauts and the astronauts communicate
with the robot?
We'll also talk about artificial intelligence.
How does that robot know what to do?
I would like to introduce our robot experts.
Dan Andrews in controls and automation.
Keith Nicewarner a robotics engineer and Sal Desiano who is a research
I'll hand it over to Dan to talk about propulsion.
>> Nice to see all of you out there, I'm Dan.
First what I'm going to talk about is propulsion.
I was impressed by what I saw submitted.
A wide range of propulsion ideas from fans and propelors, some small,
some quite large.
Some were articulating, meaning they would move around.
Others were fixed but they would put a rudder behind them to steer it
like a boat.
Thrusters were another option, compressed gas, air, sometimes combustion,
I saw one that actually worked like a jet.
Wheels, traditional wheels, castor wheels that can flop around, tank tracks,
A lot of them had legs or arms or different manipulators.
One had many, many legs.
And sometimes with suction cups or magnets on the end of them in order
to keep themselves positioned properly.
I saw one that actually proposed swimming through the Space Station environment,
the microgravity environment and a few had hybrids.
That is when you take a combination of propulsion means and mix them together.
I thought that was interesting.
Before I go into details I will talk about what you need to do when you're
defining your propulsion system for a robot.
The first thing you need to do is figure out where your robot is going
Is it supposed to be here on the floor on earth?
Is it supposed to be up in a spacecraft in orbit?
On another planet?
If in orbit is it inside the spaceship where there is air or is it outside
in the vacuum of space?
All these things affect what you're going to be doing with your robot.
An example might be wheels.
If you're up on the Space Station and you propose your robot having wheels
and there is little gravity up there how will the wheels stay attached
to the walls and floors?
Hard though do.
Maybe not a good solution there.
But what about other Biomedic solutions?
Are there things that you can pick from nature the way a Caterpillar walks,
the way a fish swims, the way humans walk?
And then there is other issues that are sort of peripheral like touch
screens if your design wishes to have the astronauts work with touch screens.
What happens when you push on a robot that is floating up on the station?
It floats away after you push the button.
What are you going to do to contend with that?
These are all things you have to think about when designing your robot.
There were, I thought, three highly notable good student proposals that
we saw the first one was north ridge middle school, Mrs. brown's class.
On this one they proposed a multi-capable robot.
An example of the hybrid I was talking about in the beginning.
They have fans for when it's operating inside the spacecraft.
So it's blowing around assuming that there is air in there and that's,
of course, a reasonable assumption.
However, it also works outside the spacecraft because they have nitrogen
gas packs in there that can work out in the vacuum of space.
And then finally, as you saw in the diagram it has arms to crawl around
the surface of a planet or whatever.
The next one I thought notable was Michael from the frog hauler school
He came up with some interesting possibilities.
This is not showing his entire robot but showing the propulsion means
which I found very interesting.
What he has blown up in his diagram there is his many, perhaps hundreds
of feet going around this robot with little suction cups on the end of
His idea there is to be able to crawl like an octopus.
Crawling around but being able to stick.
If you can't attach, you might float away.
He also proposed flippers.
I didn't see any detailed pictures to help me understand it but I like
the idea of flippers swimming through the Space Station.
Nature shows these things work.
Maybe it would work in a robot.
The K.R. Smith school.
The science club.
Mr. helper, what I found very interesting about this one is the propeller
fan system they propose using is propulsion.
They specifically mentioned having a screen over the propeller.
It sounds like a small detail but when you think about it that's a huge
detail because when this device is interacting with astronauts, their
hair is going all over.
They have fingers, they're floating.
Not necessarily totally in control of where they're going, you could have
a real problem with spinning propellers and these astronauts.
Having a screen on there is not only a good idea but required.
I thought that was worth noting.
Other schools had come up with interesting propulsion means, too.
E.M. Johnson elementary school.
The robot 1 and robot buster 1.0.
Holy cross college and interparish school came up with interesting variations
I would like to show a video now of NASA's initial attempt at a propulsion
Up on your screen you'll see now what we call the model 1 robot assistant
What this is, that's a granite table and the robot is sitting on top of
a air bearing which allows it to float with almost no friction at all.
It allows us to test navigation and everything you'll hear talked about
This variation uses fans, which many of you have proposed.
You can see the fans operating to try to control the position of the robot.
Very capable testbed for us to be able to move forward with the design.
With that, I would like to turn this over to Keith Nicewarner next to
me who will be speaking with sensors.
>> I'm Keith, a robotics engineer here at NASA.
What Dan introduced were the issues with propulsion.
So once you have a means for moving yourself around in space, the next
thing you need to do is to be able to sense your environment around you.
And what allows the entries did was recognize there are two types of sensing
that a robot needs to do.
One is, you need to be able to sense the environment such as the pressure,
the humidity, the oxygen.
To make sure that it's still safe for astronauts to breathe.
Another type of instrument are navigation centers that are used to help
the robot know where it is, know if it's about to bump into something.
Be able to move around.
So there were several that I felt were very interesting from the entries.
I liked them all.
But there is a few I want to just talk about.
One was from the UW Fritz elementary, Mrs. waltz 5th grade class.
They in particular had a GPS system for navigation which I thought was
That's something that we actually looked at.
And is used on many NASA space robots.
Spectrometer, they -- this class proposed using a spectrometer for measuring
It's a very sophisticated instrument that allows you to measure all the
chemical composition of gases and things around you.
The vision system that they proposed allowed them to use two cameras to
This is something that we use -- we looked at as well.
They also had an interesting idea for looking at night vision in case
the lights go out and you need to see where you are.
So it would be good to be able to see in the dark.
Another entry was a middle school, this is a robot they called aerotech.
This one had sonar proximity sensors.
It allows you to project infrared and read it back and be able the tell
how far away things are.
They also said -- proposed having carbon dioxide sensors on board which
is a very critical gas to measure on the Space Station because you don't
want to have a lot of carbon dioxide building up.
It's bad for the astronauts.
Another one that I thought was mentionable was the Carolyn robins elementary
school, the 7th grade.
This is a robot they call dude.
It has cameras and they have motion sensors which I thought was very interesting,
because when you're moving around inside the Space Station or outside
the Space Station, working with astronauts, the astronauts are moving,
the tools they use are moving.
Being able to detect their motion is very important.
Their design also had a thermometer so they could measure the temperature
That's important for being able to tell if there is a problem on station
if it's getting too hot, the robot can tell you.
You have a microphone so you can talk to the robot.
Some others that I'll just briefly mention.
The E.M. Johnson elementary school robot 2, Belleville junior high and
the holy cross college, their design.
All of these had very interesting sensor configurations.
So I can talk a little bit more about what we've done in our testing.
We've focused on, like I said, proximity sensors for trying to tell if
there is something nearby us but we've also been looking at how to process
Ideally we want to process vision just the same way people do, using two
What I can show you if we roll the video, this is actually two camera
views, a left and right camera.
This robot is moving around inside a Space Station mock-up that allows
us to do testing.
You'll see soon here the robot has a stereo, the information process gives
The green blobs show how far away things are.
The sensing information is very important for environment and for navigation,
for navigation and thinking about the navigation.
I'll hand it off to Sal.
>> The ultimate goal for any row -- robot to do something useful.
What the real goal is to build artificial intelligence that can actually
get the robot to do tasks and things that you don't want to do.
Its what robots are for.
You can't do the interesting thing until the robot has some abilities.
If you were to build a robot that can go to the store and buy juice and
milk and ice cream.
It didn't have any legs it wouldn't get anywhere.
He can't get to the store.
So what navigation is, it's one of those things you don't think about
until after you build the robot and thing well, all right, we have to
get this to do something.
It takes the sensors that Keith talked about and the propulsion system
that Dan talked about and ties them all together to do interesting things.
Some of the things that we have to work on show -- is basically getting
the robot to move around inside the Space Station which we can show you
an image of in this video.
This is the International Space Station, at least our computer simulation
If you really carefully look in the middle you can see a small red ball
and an astronaut with no arms or legs.
And that's what we use to test our navigation system before we put it
on the actual robot.
What our navigation system does, it does things like if you know that
you're here and you need to get to there, that's easy, you just walk,
What happens if you want to get from here to there and there are chairs
in the way.
You have to go around the chairs.
If I wanted to get back to where the camera is I'd have to go make a left,
go up the stairs, make a right, avoid the stair banister over there and
it's easy for me to say because I've been training on this for 25 years.
In robots you have to teach them to do this.
That's what the navigation systems are.
Other things you have to do is if something pops up in your way so you're
trying to get to the bathroom and your brother pops in the way and you
have to get there, the only thing you can do is get around him or move
But you should go around your brother.
The only way to do that is to realize there is something there, an obstacle
appeared plan a path around it like playing football.
We have to teach the robots to do that.
You can't say robot, don't touch anything.
You have to tell it how.
One of the interesting robots that handled this was sar 1.
And what that robot actually had is microskopic sensors on the top of
They made sure you didn't hit the sides of the station.
You don't want a robot to come over, hit a switch and open the door.
Leaving the station is bad when you're in space.
You have sensors to make sure you don't hit those.
Another design is dude, and what dude has is dude has motion sensors.
He's wandering around and it senses things that are moving around.
If you get an astronaut in front of you and it's moving, dude won't hit
Safety is important to us.
Dude makes sure when it's moving around that it doesn't hit anything.
A couple other schools that actually did really interesting things with
navigation were from Robinson junior high school and from a lock mere
Once you actually have the robot doing something with some ability.
A robot that can move around to get from here to there.
You can say go to the refrig and get me juice and the robot can do it.
How do you actually get the robot to be smart enough to figure out exactly
what it means?
That skill, that idea of figuring out how to do something is called intelligence.
You and I have it.
We were born with it.
The robots don't have it.
They have to have it put in them.
Artificial intelligence is what it's called.
There are lots of different things you can do with artificial intelligence.
If you want the robot to wander around and record sensor readings all
over the place.
It wanders by and it's 150 degrees.
If it's that hot an astronaut will burn himself.
Well, we should probably tell somebody.
So you need an robot to actually be able to figure out that entire sequence
of thoughts, which is not that easy.
So what we did is we spent a lot of time working on systems that can taken
puts from the sensors and figure out what to do with them and actually
do something with the propulsion.
One of the things we have is we're working on systems to have a laser
pointer so one of the problems we have in the Space Station is the astronauts
are doing experiments, right?
They're sitting there trying to move this thing over there and this isn't
their experiment, it's someone else's so they have to do it because they're
The astronaut realizes they don't know exactly what they are supposed
to be doing.
The scientist on the ground in Houston says take that and move it over
The astronaut says which?
He says that.
The astronaut goes this?
No left and this goes on for a while and this is expensive.
We're looking at putting a laser pointer that you have when you're doing
a presentation so you can actually have the robot point at the thing the
scientist is trying to do.
And so what that allows the scientist to do, the scientist can talk to
the astronaut through the robot.
Other things we have are the robot can wander around and make sure there
are no problems.
Look at experiments when the astronauts are doing other things or the
robot can sit there and say you put the socks on first and then you put
the shoes on.
The astronaut can follow the instructions because the astronauts do hundreds
of things every day and no way for them to remember everything.
A couple of the designs that do things with artificial intelligence.
One of them was Mr. Helper and he's actually here although he's off camera,
you can't see him.
What he does is actually has the ability to go get tools at the right
time and bring them to the astronaut when they need them.
That sounds simple.
We had to design a really complicated system.
I can show you a video of what this looks like.
What this is a whole set of things.
Each line is a different thing the system has to keep track of.
One line is experiments the astronaut is is doing.
What line is what tool the robot has and it uses the normal piece of software
to figure out when it has to get the screwdriver, get the astronaut and
tell him to fix the experiment.
Another interesting one was -- what they suggested was their robot should
actually do mack ground monitoring.
What we did.
Wandering around and making sure nothing is going on.
It doesn't have to be complicated.
It doesn't have to be some fantastic task.
Simple and useful things are fantastically useful.
Robot buster, Adam actually built a version of his and it's on this next
Out of a lot of cans.
Adam's robot actually follows rules.
I want to -- it walks but does not run.
It moves slowly and sometimes it moves fast.
It seems obvious to you.
You don't run when you're by a pool or in a hallway at school.
And you don't walk slowly when you're being chased by a bully.
Those are very simple rules.
You know them.
It took you 15 years to learn them.
For the robot we have to have the intelligence to figure all these out.
That is what the artificial intelligence is.
Taking rules we know and coding them in such a way that the robot can
actually follow them.
We don't want the robot to get detention, fall in a pool or get caught
by a pulley.
A couple other robots was bought 1,000 from lock mere academy.
You have a robot that does all these cool things.
He's not falling in the pool unless he wants to.
You can one that actually be useful.
They're not actually useful until you can tell the robot what to do.
If the robot sits there and says there are so many things I can do, you
have no idea.
You want to be able to say to the robot go to the store and get me juice.
>> Communication is an important part because, as artificial intelligence
gets smarter and smarter it becomes more and more important for us to
be able to tell what the robot is trying to do.
So if the robot is wandering by and the robot -- the astronaut is not
using the robot at that time the astronaut may want to know what are you
We want a natural way of being able to talk to the robot to be able to
understand what it's doing, what it's intent is.
And to be able to express our intent for the robot to be able to tell
it, I want you to go over to this module and check the temperature on
an experiment or something.
So telling it to do that is very natural for us to do, but it is very
difficult for robots to do this.
And I should mention some people actually in their submissions found the
-- there was some -- identified some of the problems with communication.
One of them is a middle school, they actually identified one of the classic
problems in communicating with a computer.
They said if you tell the robot to draw the curtains, then how does the
robot know it should take out a pen and start drawing or if it should
go and move the curtains?
It takes common sense and knowledge which is very difficult to put into
Another class that identified some of these problems was happy middle
school the aerotech robot.
They mentioned how difficult it is to communicate with the robot trying
to have a natural conversation with a robot.
It's much easier to communicate with a computer or robot when you have
a specific topic to talk about.
>> I have a question here from the floor with our visiting students
>> All right.
>> Do you think silicon is good to use in space?
>> Silicon is a strong material but it is very brittle and it is
also pretty expensive.
So plastics and aluminum are materials that NASA usually uses for space
because they're safe and they are strong and easy to manufacture.
It's a good question, though.
>> One more.
This is on communication.
>> We're bringing in Mr. helper.
>> Would you rather prefer voice command or to type in the command
to communicate with the robot.
>> It's a good question.
There -- it depends on what type of information you're trying to communicate.
So if you're trying to tell the robot to move out of the way or stop what
it's doing, a verbal command may be the right choice.
But if you're trying to convey a picture or something that is a drawing,
then trying to describe a drawing is very difficult even for humans to
It is much easier to draw a picture and show the picture.
So for our robots and a lot of other NASA robots they use a mixture of
both vocal commands and graphical representations.
And for texture.
Being able to read text like you do a book.
It's a good question.
And the last one I wanted to mention was from fog hollow school of Michael
from Australia designed -- if we could show the next slide.
Designed a robot that has a lot of interesting features.
One of the ones that I was interested in was how his robot was designed
to play chess and games with the astronauts.
That's an interesting point.
Astronauts are typically on the Space Station a long time.
And keeping them entertained if they just work, work, work all the time
it's not good for anybody.
So giving them some time to play is good for everyone and especially the
astronauts that work very hard.
Having a robot that can actually be part of their entertainment is a good
Some other ones I would like to mention the Carolyn Roberts elementary.
They -- a group there had designed some motivateors that can talk like
people and help motivate the astronauts in case they need motivation.
Belleville junior high the robot Bob and bark low middle school for design
Everyone had some great designs for communication.
And for all the other types.
So what we've done is we've gone through all these basic areas of the
-- of what goes into building a robot and each one of your designs really
addressed those areas very well.
So the next part is how do we put these together into a prototype?
Dan is going to talk about that next.
>> So before we get into some fun parts that I have here to show
you, I thought I would discuss first of all what a prototype is to help
you understand how you might be able to use them.
We saw many of them in the submissions that came in.
A prototype can be one of two types of categories.
One type of prototype is a form prototype.
This is one that looks like the way you would like your robot to work
but it doesn't work.
It has no functionality at all but it helps you see what this robot might
How big it will be compared to humans.
Will it be too huge or too small to be seen and so forth.
The other type of prototype is the functional prototype.
This is one that may not look at all like your robot is going to look.
But it does give you the ability to test out functionality.
Is this thing going to be too loud?
Does this thing move fast enough?
Is it able to make decisions when obstacles are put in the way?
So we've done a little of both in our own work trying to create our own
little robot assistant.
I thought I would start by showing you some of the different prototypes
that we came up with, functional prototypes, not form prototypes.
What we have here are some axial fans these are called.
They look like blow dryer fans or anything else that we had considered
for the -- our robot.
There is a huge range of these out there.
You can see the one on the right here is very small, probably easy to
package so that would be a nice one to use.
However, it is so small that it might not have sufficient thrust.
How do you know?
You have to try it and test it out.
This one in the middle a little bit bigger has more blades on it, probably
capable of more thrust.
We know it is, we tested it.
A little harder to package.
Of course, this one here has plenty of thrust.
We have found that out.
But it's big.
This is the one, by the way, that we tried out in the videotape that you
saw earlier of the robot moving on the granite table.
Other -- when you get those fans going you have to start looking at how
to test them.
So we would put them in a duct.
Here is a test duct and we have two of these smaller fans inside of here
and we would try, you know, blowing air through them with one working
in the forward direction, one in the reverse.
And then we even start experimenting with other ideas we hadn't initially
You see the pipe coming off the side here.
We're looking at the possibility of inhaling air through both of these
sides here instead of passing it through and blowing it outside ways.
Maybe that gives you additional ability to control the robot.
Then we started looking at entirely different means.
This is called a blower.
It does basically the same thing as the other fans that you saw, but it's
sort of a different configuration.
The air gets sucked in through the middle here and blown out in all directions
around the blower.
So we would design up a housing like this guy here.
You can see it's well used, beat up, cracked and all the rest.
This blower would go inside this housing.
Again, this is a functional prototype.
We're trying to see how it works.
This doesn't necessarily look right and we don't care.
We want to see how well this works.
It has multiple outlet ducts on here.
What we would do is cover up all of them for one and test what the flow
looks like and how well it will work coming out of here and then we test
It's the best way to find out what works.
This is the type of thing you students could easily do as well.
It's a basic approach that you would take on such a thing.
Then we even have some I would say more of the usual concepts.
This one is rather fragile.
Not that it will break but it's hard to keep aligned.
I call this one the bathroom fan one.
What it has are these louvers inside.
I'm not sure how well you can see them.
You can see there is three different louvers in here and the blower that
I showed you in the previous scenario fits right in here in the same way.
Then so the air comes in from the top and then blows out through these
ducts on the sides all the way around.
If you can change the shape of these louvers, you can change where the
air is blowing and therefore change what your robot is doing.
So I have some images here of these devices undergoing tests.
This first still that you see shows the axial fans in that duct.
They're actually attached to a force transducer which tells us how much
thrust we're getting out of the duct.
Gives us a way to evaluate it.
Here is another picture up on the top.
This is the blower you just saw.
We have the blower inside of the housing and in this case we're testing
the duct that is down here blowing down.
And so the force transducer can tell us how much thrust we're getting
compared to how much power we're putting into it.
It helps us evaluate if this is a good idea or not so good an idea.
This one I didn't have the prototype here with me.
This was another consideration.
Sort of a helicopter designed where you have propellers along the top.
This is something that would be very good if we expected this robot to
work here on earth.
You can get a lot of uplifting force like you get with a helicopter and
allow it to buzz around this room.
It turned out to have some complications with respect to using it on a
Loud, uses a lot of power.
So therefore it is not really very desirable for the use we intended it.
I think it's -- I'm sorry, go to the video.
We've got some video of our blower undergoing tests.
This is our whole test rig here.
You can see obviously where the air is going.
It's being sucked in here, going through the blower and blowing out the
This whole rig it's sitting on is measuring how much force is coming out
This is a repeat with a different camera angle.
On this big thing back here, look at the bar coming out of the can.
It is showing how much torque is coming out of the motor when it's operating.
Watch it moves.
As the motor slows down, see how it jerks?
It will make your whole robot jerk.
You need to look at how your robot will handle the control system or maybe
you don't change the speed very fast.
Sometimes prototyping is done in software.
It helps you see what is going on.
You don't have to build anything and it can help you go through all the
possibilities very quickly.
So I would like to bring up Robinson junior high's submission here.
I thought this was worth noting because the 8th graders proposed an interesting
thrust-directing approach with their propulsion.
A hot wheels crossroad is how they described it.
What they're trying to do here is direct flow in a way they would like
it to be directed in order to steer the robot.
That's exactly what many of our considerations were looking at.
There are a number of ways to do it.
Many of the student groups came up with lots of alternatives.
I found this quite interesting.
These work but don't necessarily look like anything that means anything
The other category is the form prototype.
This is the category that doesn't work, but it helps you see what it is
that you're building.
Here I would like to go to the next slide.
>> Could I ask a question here about materials in building these?
I have a couple of those questions.
Michael has chimed in from TAZMANIA about aluminum.
In his experience it's rather soft.
He's seen it around the house for window frames and apparently very bent
window screens, fly screens, is there anything special you can do to make
it more suitable for planes and space robots?
>> That is a good question.
When you're comparing things like windows and screens and so forth, those
are generally designed to keep bugs out and the like and are not seeing
any real structural load, if you will.
When you go and head and use aluminum in a shell on a space robot or anything.
If you make it sufficiently thick, what nice about aluminum it's lightweight.
It is a good solution.
When you test here on earth when we're doing the testing of our robot
here, making aluminum might make it heavy which would be fine on the Space
Station but a problem here.
That's why you see many of our prototypes made out of plastic because
it's fairly strong and it's very light.
Any other questions?
>> There are actually.
I'm not sure whether you want to interrupt at this moment for them.
But one of the people, I'm sorry they didn't identify themselves.
How long does it take to make a robot?
>> How long has it taken us so far?
Define what your robot is going to do and all the other I cany stuff is
how much money do you have.
What is the calendar.
When does the thing have to be launched?
A whole bunch of other stuff that don't have to do with the technical
I don't think there is a specific answer to that but it's a good question
to ask on the particular robot you're working on.
You can turn around prototypes like I'm describing very quickly.
It doesn't mean they're ready to be deployed.
It is an educational tool to help you get smarter in building the final
device that will be launched.
Up on the screen now what you'll see are some of our solid model renderings
of different shell configurations.
We started looking at a sphere.
We like the idea not having any sharp corners that might hurt astronauts
and a number of other reasons.
And the top left corner you see a blower kind of like the blower I held
That is one that is actually in the shell.
So the top right and the bottom left are showing axial fans.
See the ducts in there?
It's like the ducts I was holding up.
Then you see on the bottom right ones that have little side ports like
the little test fixture I was talking about.
We were playing with the idea of using that as propulsion.
Here are some more idea.
What is fun about this part here.
Because it's a software model you can get as crazy an idea as to wish
to explore it which leads to other ideas.
This one is a BUBBA jet.
It has a giant fan all the way around the outside of it which is actually
a smart thing to do.
It has other limitations so we haven't chosen that, but it's an intrigue
-- an intriguing option.
On this front with respect to software tools I thought it was worth noting
bark lo middle school used software solid modeling tools that were designed
for children that I had personally never seen before but I found it to
be a great leverage that they could have.
You don't have to have the software tools to do this.
Pencil and paper works fine.
What we found is by being able to model this on the computer we can quickly
go through a lot of different options and variations and start to mature
a design towards something that is plausible.
It gets us to what NASA is doing.
What does our robot look like?
Pull up the first slide.
This is our first prototype.
We call this robot the Personal Satellite Assistant.
Or PSA for short.
Some of you may have seen this on other shows or the web.
This is a form prototype.
This thing doesn't exist as a working robot but it helped us to see what
size should this be?
What types of things might have to be poking through the shell to do what
we need to do?
Can we fit all those on the shell, and so forth?
That's our first prototype.
The next slide, please.
This one should be familiar.
The one you saw on the video.
This is what we call our model 1PSA.
This PSA is a functional prototype.
It helps us see how our design is going, helps us code the software that
we need to put within it so that it can do collision avoidance, stereo
All the stuff we need to do.
Doesn't look like a ball.
Not intended to.
This was our first attempt to do both.
This is a form and function prototype.
So it sort of looks like the pictures that you've been seeing like a prototype
but it actually has all the stuff of the model one inside of it.
In order to compromise that, it means this is large.
This is not, you know, a small thing.
This is closer to a basketball size device, 12 inches in diameter but
it does work and what you see it sitting on there is an air bearing sand.
It can run around on the granite table.
We have video of this thing running within a crane structure that we built
that is intending to emulate the Space Station.
The crane itself makes for a microgravity environment.
Surrounding it is some Space Station rock mock-up's that look at PSA,
see what it's likely to see on station or a close approximation.
So that is sort of an upview.
You can see in the very bottom of the screen the PSA in the previous shot.
It is hanging from these yellow cords.
Lots of instruments and electronics to sense what it is doing.
The whole point is to sense what the ball is doing so that the ball doesn't
even know that it isn't out in space and flying around.
So there is some of the electronics getting adjusted.
A whole bunch of things going on there.
So here it is flying with the axial fans, remember the fans in the ducts?
Flying through the Space Station mock-up.
This isn't up there but this is over in a building here at Ames Research
See the fans operating very similarly to that model one functional mock-up
that I showed?
It's a very similar approach.
It is just packaged now.
So that then leads us to the model 3 mock-up which I'm holding in front
of me here.
It is the same size as the model 2 that you just saw in the video.
It has a little different design.
We went ahead and are working with a blower design up here similar to
my second functional prototypes with exit vents.
This is a functional and form prototype.
This will go onto the crane structure that you saw up there and be the
next evolution of the PSA.
The reason we went to something like this was because the model 2 PSA
with the axial fans.
If we ever expect this to get small, those fans will have to get ridiculously
small and then they'll have to spin at unbelievable speeds to get the
thrust you need out of them.
So in this business we call that a dead end.
It isn't going to go anywhere once we shrink it down.
We had to think of a new way.
What can shrink when we get there?
This is where we currently believe we can go.
The blowers can shrink down nicely and we should be able to accommodate
it in a smaller ball which leads us to model 4.
Go ahead and put this stand down.
This is, once again, a form mock-up.
This doesn't work.
Much like the first one.
Not intended to.
This is a visual representation of what the PSA model 4 would look like.
It's about the size we would like to see it but we don't presently have
the ability to put in the electronics at this density to get it to work.
That's where we're heading.
You can see inlet ports.
Let me get this right for you.
Inlet ports on the side that are over here.
Right over here and over here on both sides of the PSA.
That's where the air gets sucked in and then it gets blown out through
the top ports.
There are four on the top and four on the bottom.
By controlling those in the louvers you saw in the previous mock-up this
thing can fly up, down and around.
Keith will be talking about some of the sensors on the device.
>> So this, like some of our other prototypes, has a lot of sensors,
both environmental and navigation.
The obvious ones are the two -- we have cameras here and here.
We also have cameras on the sides and in the back.
So we have cameras looking all different directions which is unique.
Our robot has eyes in the back of its head.
It allows it to see what is in front of it and behind it and on the sides.
So it also has micro phone for listening to things in the environment.
It has another special camera for teleconferenceing that allows it to
an astronaut that is is floating in front of the PSA can communicate with
someone on the ground.
Another interesting thing is the LCD in the front which allows us to display
graphical or video information or text information to the astronauts.
This is very important for doing -- a lot of the astronaut's job is communicating
with people on the ground.
Another thing this robot has is a thermal camera that allows it to look
for hot spots.
This is very important for doing things like looking for an overheating
rack or things like that so the robot can do that.
And then I'll hand it to Sal and he can talk about some of the other things.
>> So remember before I told you the naiv -- navigation is taking
the parts and making it useful.
You won't see most of them.
They're inside in the computer.
Most robots have computers inside.
But I do want to point out that a couple of things we have are, we have
the blowers that Dan pointed out because we use that to move around.
We have the sonars.
If you look over here there are these little dot holes.
They're all over the place.
They're here and they're here.
Now I know what a weather man feels like.
What they do is they're the rain sensors.
They let you know that you are going to run into something and then you
We also have the cameras.
We also use them for figuring out how far away things are.
If you have two eyes, most of us do, you can see -- well, if you have
one eye you can't tell how far away things are.
Have you tried walking down a hall with one eye covered?
It's a bad idea.
You'll hit the wall.
If you have two ice -- eyes you can do that.
One of the neat things about this robot and one of the ways we get it
so small is the cameras and the screen and all these different parts are
used for multiple parts of the system.
So the camera is used for teleconferenceing and navigation.
The computer is used for vision and artificial intelligence.
That way we can get everything in this really small ball, we hope.
The other section I wanted to talk about is artificial intelligence.
It again is mostly inside in the computer.
The neat thing is that because all the robots that we built have the same
computer inside we can actually use the same software inside.
The same artificial intelligence that worked on the flat one, the big
metal box, actually works on the big round one and hopefully Sunday will
work on the little round one.
I want to mention one trick that we use that nobody thought of.
I'm pretty proud of it.
We have what they call wireless ethernet in here.
You know how when you plug your computer into the wall it can talk to
We have a thing in here that can talk to other computers without being
plugged into anything.
We have a small computer in here but then a huge computer in the next
room that can do high level thinking.
We have a small robot with a big brain.
We don't have to worry about fitting it in there.
It's how we do the artificial intelligence.
>> I have a question that has to do with just this topic.
Students from Robinson junior high in Robinson, Texas.
We read an expert's advice to use compact flash drive instead of hard
drive that spins.
The reason being that the spinning hard drive would make the robot spin
in the opposite direction.
How do you keep a spinning fan from doing the same thing?
Way to catch me on mistakes.
So -- wow.
>> That is--
>> There is a good answer to that.
There is two answers.
I forgot to mention, if you read the expert's response to your preliminary
designs, that was me.
I take responsibility for anything I might have said.
I had a lot of fun reading a lot of your designs, particularly the peppered
I can't remember their entire name.
It's probably 150 letters long.
The way we do it actually and Dan's team for engineering did a really
good job of this.
You see how there are fans on the top and the bottom.
Turns out that if you spin one set of fans one way an the other set the
other way, that it cancels out.
So if you can actually sit here.
If you actually want the robot to spin, you spin them both the same way
and the robot spins on purpose.
The answer is, you actually have to deal with it with fans.
You have to make sure you have one fan spinning one way and one the other
way or for some reason before it was spinning so when you turn on the
fan it stops.
The other reason we did that was because the hard drives actually break
when they're in robots.
If you ever take your computer and I don't recommend you do this because
it's your parents and you shake it, eventually it is not going to remember
Everything on the hard drive is gone.
We had the same problem with our robots.
We would build a robot and had a hard drive in it like you have in your
computer and it shakes when the fan turns on.
Every two or three weeks the robot would forget everything and we would
have to buy a new hard drive.
It's why I suggested the compact flash.
>> We have a couple of other questions unless you're still continuing,
Here is a question that says, how much do robots usually cost?
>> The answer is, how much can you afford to pay?
We have a system here, whoever has the robot gets to talk.
I have the robot right now.
Let's talk about money depending on how complicated they are.
One of the rob -- robots the big red ball was $10,000.
You can buy a robot on the web.
They have kids you can buy now, I wish I remember the name of them for
students and kids and you can actually build your own robot and program
it and they only cost like $150.
Which may seem like a lot of money to you but for me and NASA, that's
>> The majority of the cost for robots is the development.
The actual materials don't cost that much.
But the amount of time that us engineers and scientists spend on the robot
are a lot more expensive than the actual materials for it.
>> I have to think about it.
You don't think about it but we have to eat.
>> We have to get paid.
>> In order for us to eat we have to get paid.
You have to pay us for our food that's where a lot of the costs for robots
>> One of my favorite questions also.
I believe this came from a math class in Columbia, Maryland.
Somebody asks, how can I become a robot engineer?
Study hard and build robots.
>> I would say it's very important.
I'm the prototype guy.
It's very important to play.
If you are interested in robotics, go out there and try stuff.
It doesn't matter how silly it is or how complex it is.
The first things you do will be silly and simple and all that.
Then you'll learn from it and try something else and it will be kind of
Then you'll go further and back it up with some education.
Going to school, learning practical experience and before you know it
you're playing in building your own robots.
>> All of us have that same background.
We were always building things and we built earlier things that didn't
work too well and as you learn, you realize what you can learn from school
can be applied to the problems that you're trying to build and you just
keep building on that and pretty soon you're building robots.
>> Just to add to that.
I don't know we were all always building things.
I started my career in robotics when I was very young mainly breaking
I spent a good deal of my childhood taking things apart.
Took apart my parents' hair dryer, the TV when we replaced that.
Only when I was 16 or 17 that I was actually able to put anything back
I was trying to take stuff apart, filing our out how it worked.
Build my own things.
Take courses and I would actually take whatever opportunity I had.
If you have a club.
One of the schools had a robots club.
It was great.
You actually try building robots.
When you have our job you say I tried that once and it didn't work so
we'll do it this other way.
It is exactly how you do robots.
Do as many as you can and learn everything you can from the ones you tried.
>> I promised to give you guys a chance to give me one sentence
on what is next.
What is going on next in your robot lives?
>> Well, next I guess I'm going to lunch.
But for the next year or so what I'll be working on is trying to get the
robot to move around faster.
Right now it moves kind of slow.
And still not slam into things and still do useful things while being
as small as this thing will be someday.
Right now it's very large and it has a big brain.
I want to get something really small that can move around really quick
and be just as smart.
I'll give it to Keith so he can talk.
>> Some of the things I'll be doing over the next year are I mentioned
we have all these cameras we're looking at having eight or nine cameras
on this robot.
Being able to process all those cameras at the same time is pretty challenging.
That's one of the things we'll be working on.
Another thing is the higher level intelligence.
The artificial intelligence.
We'll be trying to make it smarter and smarter as time goes by.
>> I like how we're following the bouncing red ball.
What is ahead for us is pretty obvious, in my mind.
You may recall I described this as a form prototype which is true.
We need to start to pursue how we take this device down here, this big
one that we believe we can fit everything we need in it, and get it towards
here so this is a functional prototype.
A lot of work, a lot of electronics has to be shrunk down and become high
The blowers have to be made smaller.
Everything has to shrink.
It's the next step in the evolution.
This is a form.
We want to make it a functional as well.
It's what we'll be working on in the coming year.
>> Thank you very much Dan, Keith and Sal and thank you very much
for joining us today and for entering our design challenge.
I'm going to tell you how you can find out more about the P.S.A.
Go to our website.
And to go to the next page, the next slide, You'll see a picture of the
P.S.A. and you can click on it like you can click on the camera and see
an enlarged view of that particular part of the P.S.A. and information
on it so you can find out all sorts of things about the P.S.A. and keep
going back to the website because we'll have a lot of interesting activities
on it for you soon.
So thank you very much and goodbye.