Brianna O’Neal, University of Iowa
Shuokai Chang, UC
Department of Mechanical and Nuclear Engineering
Westinghouse Fellows Program
The Pennsylvania State
University Park, PA 16802 USA
Currently no nuclear power plants are constructed around robotic systems because they are made for
functionality to humans. There are many benefits of a robotic system that could perform inspection and
maintenance tasks within an industrial setting such as the decrease of man labor. Designs of existing
robots for nuclear reactors are limited and highly complex. This pape
r concentrates on creating low cost
robotic system prototypes that are focused prima
rily on mobility.
Specifically wall climbing magnetic and
rail prototypes were chosen to serve the purpose of adhering to a given surface, such as the steel
containment of the nuclear reactor.
The magnetic robot is able to have more mobility along any
omagnetic surfaces; whereas the rail robot provides more mobility in any direction with easy
transfer between different planes of movement. More work still needs to be completed before any
realistic prototype can be implemented into a nuclear power plant.
Future work consists of creating
prototypes that are to scale
for industry and
to respond to disaster response. More issues to
research further consists of radiation effects and equipment malfunctions.
Table of Contents
Table of Contents……………………………………………………………………………………………………………3
Analysis of MagneticBot
Conclusions and F
he US Nuclear Regulatory Commission (NRC) requires that people who work with radioactive materials
can be exposed to an annual radiation dose of no more than 5 rem
According the NRC’s EPRI
(Electrical Power Research Institute) survey about the tradeoffs between human and robotic inspection
the use of robotics saves between $100,000 and $1 million
A wide range of robots
is already used by the nucl
ear industry for inspection and maintenance tasks
, yet the deployment of
robots as an integral part of the functioning of a reactor is not yet commonplace
designs of robot
can be applied in
a nuclear reactor
with a focus on
integrating the robot function with the reactor’s physical design
. Several different styles of robots were
, with particular emphasis on
tasks of inspection and mainten
There are just over 440 commercial nuclear power reactors
around the world
currently in operation
However, most of the reactors designs did not incorporate the use of robotics, mostly because
robotics is a fairly new topic
prior to the design of most current plants
Implementing robotic systems
into the nuclear reactor is proving difficult to most operato
Reactors are typically designed based on these three categories: structures whose failure could
contribute to a nuclear incident (containments), structures essential to the power operation, and
structures convenient to the operation but not nec
essarily to the safety
. Thus, the equipment and
configurations of the nuclear reactor are not designed for the ease of installation of robotic systems
rather a segregation of
containment, essential operation, and ease of use.
segregation requires that each portion of a reactor be highly optimized for it
s design task, and thus the
robots used at each location have been similarly optimized.
implies poor reus
ability and high costs.
Thus, the ability for robots to operate across
such as inspection and basic maintenance
An increase in mobility can improve a robot’s ability to perform inspection and maintenance tasks
A new challenge
is to design the nuclear reactor
environment to support
Not only can such systems
ve as surrogates for humans when
environment does not permit a human presence
To achieve high mobility,
all the surfaces
inside of a
nuclear power plant.
For example, t
he nuclear reactor core i
s already problematic in terms of a working
surface: it is
housed in a steel containment
, which is further confined by a concrete structure for
Some tasks given to robots include the inspection of pipes, cleaning various
surfaces, taking pictures
, and performing maintenance on
items such as
the fuel rods
in a nuclear
, a robot that has to inspect
a pipe will be compact and able to grasp the pipe with
some type of arm manipulators. The
drive system may need to be
flexible if there are a lot of bends in
Some robots, for example a
may have to travel over a large area
many elevation changes
And of course, robots must be able
to operate around and yet not impede
humans in their vicinity.
Many of the issues of mobility within a factory environment are solved if one moves the robot
systems away from pathways required for human motion. For example, a
design that is
climb walls with
some type of mechanism that adheres to the surface
can easily avoid walkways,
stairways, etc that conflict with human use.
are already sold of
a suction cup
designs of robots have been
considered for the inspection and
of nuclear reactors
magnetic robots, monorail
robots, wall climbing robots, and flying/hovering
Like humans, robots are exposed to radiation in a nuclear reactor environment.
Once the robot is
exposed to radiation, i
t is important to contemplate
how reliable the equipment will become, ho
critical is the failure, how long will the equipment
le is the material of the robot.
Three major areas that are concerned when exposed to radiation are the drives (gearboxes and
bearings), the sensors
(visual and audio), and the co
mmunication devices (cables and wires). The
lubrication of the bearings
and gear boxes
harden when exposed to radiation causing degradation
electronics. With enough exposure, the motors can also stop working and could
the robot to stop moving. The sensors allow a robot to detect obstacles and familiarize
itself with the environment. Distance sensors will begin to have degraded measurements with more
, while the optical sensors will have a very low qua
lity and become unreliable. Communication
cables between the operator and robot are extremely important because the robot would not know
what to do it they malfunctioned
. Radiation can cause the insulation on the cables and wires degrade
making them unrel
While the budget of this project is too limited to consider
protective measures for robotics,
consider when designing and picking
materials for a robot
in a nuclear plant environment
Most of the robots that are currently in use in nuc
lear power plants
are designed for very specific tasks,
and many platforms have no functionality beyond those specific tasks. These robots often require
manual transportation and installation at the inspection site because the structure of power plants doe
not allow much mobility. One of the main objectives of the projects detailed below was to explore the
feasibility of certain robot systems designs
in a nuclear plant environment
aspect that required more consideration was vertical motion, which is a major challenge. The ability to
climb walls greatly increases the areas that robots can access
, but brings great design challenges to the
he robot and the surface the robot must use for motion
considered focusing primarily on achieving the simplest task of
inspection. The major areas examined were as follows: reliability, constr
radiation effects, complexity of the design, size, and expense. If any of these restrictions were not fully
met or impossible, the design was disregarded.
guided robot was omitted because the equipment woul
unreliable as the cable was in tension and could become tangled
once inside the nuclear reactor,
The magnetic robot was the next robot scrutinized and was one of the highest contenders to build a
prototype. The magnets were relatively permanent since they able to stick to the steel containment and
the assembly or installation of the robot would be relatively easy. Also, the size could be compact and
A suction or vacuum wa
ll climbing robot was another design considered for the prototypes. However,
the complexity of design made it fall to bottom of the contenders. According
t was also marked with high points because the rails could be pre
t into the construction
of the nuclear reactor power plant. The complexity of the design was also reasonable to complete with
the time constraints of the project.
g robots were omitted because the design would be very difficult to demonstrate. They
seemed unreliable since the flying mechanism could fail and cause the robot to fall into unwanted
Bearing in mind the multiple designs of robots, the magnetic and monorail robot were chosen
based on the criteria for inspe
ction robots. Robotic prototypes were constructed to
demonstrate how these two designs compare to robots already in existence. Different parameters were
considered in order to obtain the desired results of inspection robots. The end goal was to demonstrat
to Westinghouse the use of magnetic and monorail robot prototypes in a non
and how they compare to the existing robots in nuclear reactors. Additionally, the resulting date will
show all the constraints that go into designing a robo
t for radiation environments.
Methodology: Prototype Testing
From the idea generation, the magnetic and rail robots were chosen because they received the highest
marks and the complexity of the designs could be simulated in the lab available.
All the prototypes
an embedded system, the
, as shown in Figure
Arduino MEGA communication device
Arduinos are simple
microcontrollers and can be used to control motors, lights, and
outputs. They can be powered a 9V battery or USB computer cable.
This processor was chosen because
it was readily available to use and easily programmable. The Arduino
was fairly small, so the robot could
be easily designed around it.
It was also relatively inexpensive
for the budget to build a prototype
the robot prototypes
, the Arudino
is used to
the drive motors of the robot, which for the
prototypes that follow are hacked versions of hobby
Servos are a sma
ll electric motor with gears tha
t can turn an object, like a wheel or robotic arm. Wires
were connected from the servo to the Arduino board for proper communication between them. The
black wire is ground and the red and white wires are for power(5V) and t
he signal, respectively.
2, a simple hobby servo is displayed.
Servo with communication wires
These simple hobby servos were also readily available and
programming codes can be
downloaded from the Arduino website
for simple movements of the servos. The codes used for the
prototypes can be found in Appendix A.
e magnetic robot
, hereafter referred to as
magnets to adhere to ferrous
a high contender when considering the
other prototypes for inspection robots. As mentioned above in the Idea Generation Section, there was
a robot must meet in order to achieve the task of inspection
the first one
. By using rare earth NdFeB (Neodymium Iron Boron) magnets the design bec
permanent compared to the other samples. NdFeB magnets have excellent magnetic properties to
generate high powerful magnetic forces
Based on this kn
owledge, four NdFeB magnets were chosen
to initiate the design of the
stored in a
steel containment, the
magnets will attract
to steel. From the available
materials provided, 1.5” diameter steel wheels were chosen and cut to a thickness of .25”.
Two of the
wheels were going to be connected to hobby servos to provide the driving
while the other
two were going to be on
bearing system. Thus, quarter inch holes needed to be drilled
steel wheels so the servos could be attached and the axel could be slid through the wheel.
Using the dimensions of the steel wh
eels, the magnets were chosen to fit appropri
ately. Two NdFeB N52
magnets, with a diameter of 1” and thickness of .125”, were selected to fit on the steel wheels that were
going to be attached to the servos.
As seen by the proportions, NdFeB N52 magnets help reduce the
weight of the wheel
other two magnets with an outer diameter of 1”, inner diameter of .25”
and thickness of .25” were then glided on to the axels. These magnets were called NdFeB N45 and were
relatively inexpensive, while fitting the parameters of
d proportions. Fig
magnets used for this robot prototype.
Rare Earth Magnets N54 and N45, respectively
Using both of these magnets and the 16 gauge steel, the magnet force could be calculated. The force
was calculated because it helped reinsure how much th magnets could hold including the weight o
robot. In Figure 4 and 5,
a schematic of the pull for
ce is displayed.
Pull force dependent between the steel plate and N52,
Judging from Figure 4, the N45 magnet will be have
a stronger pull force. In Figure 5, a graph is displayed
of the expected pull force.
Expected pull force of N52 and N45, respectively
The predictions proved right because N52 had an e
xpected pull force of 14.40lb; whereas the N45
magnet had an expected full force of 21.75lb. Therefore with four total magnets the entire pull force is
72.3 lb, a very strong hold.
However, in this case a strong hold can be a disadvantage because it will
cause a robot to stall. Testing was then done to see how much torque was needed to make these
magnets turn. T
wo steel wheels were placed on either side of
the magnets to ensure sta
bility and to
help protect the magnet from slipping.
The magnet polarization is axial, meaning the flux is closed
through the rims of the wheel and the contact surface of the wheel.
This then provides a significant
Force v. Plate Thickness
Force v. Plate Thickness
magnetic force on the wheels.
demonstrates the powerful magnets on the steel
Steel wheels attached to magnets
ter the wheels were assembled, the design of the body was able to be envisioned.
ize of the
robot was another constraint that was cons
idered when constructing the body. The bigger the robot
becomes, the harder it is for a robot to turn.
This means large robots will be unable to fit into confined
spaces that might be essential for inspection and maintenance. Also a higher amount of torq
ue for a
large robot is necessary to guarantee that it will be able to move.
Keeping the size small, the design of the robot was relatively modest. Thus, the design consisted of four
sides and a base that fit into each other
similar to puzzle pieces.
, the robot was
assembled by creating
is important to choose the material that the robot will be made of before all the portions can be
joined together because every piece will have the s
In this case, the body of the robots
was comprised of McMaster’s
, 0.25” thick acrylic.
Expending this dimension, notches of
0.25” were cut in or through the pieces so they could be easily assembled. In Figures
dividual sections of the robot are shown with the manufactured dimensions.
Sides of the Robot
made of Acrylic
The hole with the diameter of 0.625” i
s designed for a bearing with an inner diameter of 0.25”, in which
will slide through. The other end of the side of the robot was extruded so a servo could fit
through the slot.
Referring to Figure 5
, a 0.25” slot was cut so it could slide into the base, which is shown
Base of the robot with no
tches to fit the other fragments of the robot
To complete the body of the robot, two additional sides were designed to prevent the Arduino and
battery pack from falling off
as shown in Figure
Side of Robot
of the MagneticBot 1.0
were completed, t
he SolidWorks drawings were then
sent to a laser cutter to form the appropriate parts.
All the pieces of acrylic were then assembled to
form the completed
as shown in Figure
Analysis of MagneticBot 1.0
After the mechanics of the robot
and some minor
of the designs were
coding of the robot could begin.
The Arduino was programmed to make the servos move in a continuous locomotion, meaning the robot
would travel straight for any desired time.
The robot was able to travel straight horizontally and at times
, but it was found that the driving
surface had to be completely smooth. Since the magnets
o powerful, the robot could
on any ferromagnetic surface (steel) upside down.
Although the robot could adhere in any direction, the servos did not have enough torque to overcome
the magnetic forces
to turn the robot in any way other than straight horizontally
o code was conducted to make the robot turn while on a horizontal surface and again the servos
could not maintain enough friction against the surface.
Adjusting problems to form MagneticBot
To increase the ability of the robot to drive vertically and
to spin in place, n
servos with higher torque
ordered. To ensure that the servos would have enough torque to overpower the magnetic
as well as their own weight, the servos were sized to have the ability to apply a force equivalent to
factor of three
may be significantly
oversized, but any extra force simply enables additional capability to carry large payloads.
because the robot needed to be able carry it’s
extra force because the
magnets created a lot of friction
The torque (T)
of the robot was calculated by the following equation:
T= F x r
where F, the force of the servo
r is the radius of the steel wheel (0.75”).
Assuming that force must
be sufficient to overcome the weight of the robot times some multiple k,
where m is mass, and g Is the gravitational const
ant, r is the radius
, and T is the torque available from
For the 180.5 oz/in torque servo, t
he obtained solution for equation 2 was approximately 3.5,
the satisfied the constraints of overcoming the magnetic force.
needed because the large distance between the axl
. Any protrusions away from the drive wheels
a restriction on the angles that the
robot could maneuver over
Also the corners/edges of the first prototype obstruct
ed the robot from
, as shown in Figure
MagneticBot 1.0 is restricted by a specific radius
The corners of the
design greatly prohibited the robot from turning over arcs
or bends that may be
needed to join one wall to an
other, or for the robot to navigate over a pipe
. According to Figure 9, the
radius of circle must be 9.13” or bigger for the robot to be able to move over a curve.
This is a significant
especially in a nuclear reactor
of the large numbers
It was also predicted that
wheel design could still require large torques in order to turn the
robot in place, due to
the large magnetic force from the four magnets.
a new prototype
was designed to correct
these problems. T
he new design would incorporate only two magnetic wheels
and reducing the size of the walls to a minimum
prototype was made more compact to be able to travel into smaller spaces. This then
number of wheels to two. The same magnetic wheels that were attached to the servos in MagneticBot
1.0 were reused again. However, the two steel wheels were stabilized together further by using screws
as shown in Figure
This also helped the
magnet stay in place.
The wheels were also coated with a
plastic to help provide a high friction coefficient and prevent slippage.
steadied with screws
To prevent problems of the first prototype, the high speed 180.5 oz
/in torque servos were attached to
arms were also introduced
to this robot
to help prevent it
from falling off of the edge of
steel. The arms were spring loaded so the robot was free to move over various cu
rves. This prototype
was tested to make sure it was able to fit
compared to the first
was drawn to calculate the maximum radius the robot was able to travel over, as shown in Figure
circles showing the maximum arc to be traveled
From the calcuations,t
he body of the robot was restricted to a radius of 2.86”
or smaller, whereas the
arms were restricted to a radius of 6.28”
in the upmost right position
Because the arms were spring
he arms were able to
to most surfaces with various radii
imensions of the body had to be carefully considered in order to make the robot as compact as
possible. In Figure
the side of the robot is shown. It is important to note that t
he corners were
curved to prevent them from rubbing against the surface of the steel.
The curved sides of the second
prototype were constructed first.
The side of the second
The servo dimensions are the same
as the previous robot designs,
except for the
offset by 0.08”. This
made it easier for the servo be inserted and removed easily. By
referring to Figure 12, it can be seen that the servo rectangle is not centered. Nevertheles
s, once the
wheels were attached to the servos, the wheels were then in the center of the robot. Also, t
rectangles were designed so a zip tie could be slipped through the hole to hold the Arduino and battery
in place on top of the base.
was the most difficult piece to construct because of all the restrictions. The arms had to be
placed 0.25” apart so they could encounter each other at the slots and a screw could be placed between
them. Also because the arms were allowed to rotate as the
robot interacted with different
environments, the space underneath them was limited. Consequently, the Arduino could not be placed
directly below the
arms due to the height of the Arduino itself. It then had to be placed on one side of
the base with the b
attery on the other side. This created an unusual shaped base board as shown in
The base of the robot
The small rectangle cut in the base is for a support side to hold the robot together as shown in Figure
Also the suppor
t side is connected to each of the sides as seen in Figure
Support side of the robot
As mentioned above, the slots of
arms were offset
in the base
, but they were also cut to
provide a lot of movement to the arms. They able to move 1” while rotating around a pivot point, which
was inserted into the nook (0.050”) of the base.
shows the piece of acrylic that
the arms were
attached to and we
re allowed to pivot around.
Edges that hold the arms in place
The holes in these edges match up with the holes in the arms to allow for smoo
th rotation. Below in
the arms are shown with the apporpriate dimensions.
Arms of MagneticBot 2.0
The slots of the arm are consistent with the amount of rotation
they can experience as the robot comes
across different terrains. All of the portions of the acrylic were then assembled and put together
s shown in Figure
Assembly of the second prototype
Analysis of MagneticBot 2.0
This design performed significantly better that the first design
ecause it could accommodate to more
surfaces. The arms were able to provide stability to the robot while preventing the robot from falling off
the edge of
the sides. In Figure 20
below it shows the curves that the robot was able to accomplish.
MagneticBot 2.0 was able to travel over curves
This robot was able to travel over the arcs
with an average radius of
in Figure 20
Unlike the first
prototype, this robot was
able to move while
It was also able to turn on smooth surface
in all directions.
Over some arcs and welds, the robot would tend to rotate in either the right or left
direction because one wheel would lift off the steel creating the arms to push down further on the
surface. This is because the magnetic force dec
reases more than almost 50 percent when the wheel is
tilted only 15 degrees
Hence, the high speed torque servos appeared adequate and achieved the
desired results of traveling vertically and being able to turn.
The robot was able to hold
was tested by a spring force gage. After ____ lbs, the robot started to show some deformation and then
at ____ lbs, the robot was unable to adhere to the steel
A battery pack of approximately
also employed to provide extra power to t
he servos. This ultimately made the robot travel at full speed.
The robot was able to travel at approximately _
The expected battery life is about an hour
before the servos start malfunctioning.
program to make the robot move straight and turn
can be found in Appendix A.
This code was
relatively simple and easy to replicate; however, with more complexity the robot might be able to
accomplish more tasks, such as turning around a round track.
Shortcomings in Prototype 2.0
This second prototype proved more successful than the first; however improvements can be made in
order to get the prototypes to a point where they can be tested in a nuclear reactor.
design that is pliable will
significantly improve the mobil
ity of the robot.
Both Magnetic prototypes need
able to bend because magnetic
obots that are currently in existence are “shrimp
like” or flexible
This is because the robot may have to travel up 90 degree angles creating a need for two drive joints
(one on the ground and one on a side wall as it is climbing). There are designs that incorporate bending
wheels as opposed to just a bending body
Another aspect to focus on would
components or highly resistive
erials. For instance, things like polyvinyl chloride (PVC) are not compatible for environments where
radiation is present
Error! Bookmark not defined.
]. Specifically for the magnetic rob
ot, different types
of magnets shoul
d be carefully chosen based on what procedures the robot must perform
longevity in radiation environments
. As seen by the prototypes, if the magnet force is too strong, it will
prohibit the robot to move. Fur
thermore, if the magnets are placed into environments where the
radiation is very strong, studies should be conducted to see how the magnet chosen will react.
It is also important to state that simple hobby servos would not be imple
mented into the nuclear
reactor because they are too prone to failure particularly if the magnet is too strong. Different drive
motors and larger batteries should be utilized before it can be put inside the nuclear power plant.
Using some of these suggested ideas for future pro
totypes, the robots could be tested in different
simulations to certify that it will work properly in the nuclear reactor.
as pipes and
walls would help test the prototypes to see if
they could complete the desired responsibilities.
focus of th
is paper now shifts to the a rail prototype that had similar aspects to the magnetic
based robotic systems have been implemented in manufacturing and in nuclear power
However, many of these
have very limited mobility. Rails are inherently good for horizontal
motion, but few rail
based vertical motion systems
exist. The goals of the process detailed below
were to design a robot that could travel along rails in any orientation and could transition between
different planes and directions of motion.
The first step in the design was to develop a mechanism for att
aching the robot to the rails. The initial
rails were built from 1” wide flat aluminum bars attached by bolts at regular intervals. Flanged rollers
were used. The simple prototype shown below in Figure 18 consists of two rollers held together by a
The rollers flanked the rail, and the motion of the assembly was constrained to be along the
: Roller Assembly
The initial design for a robot frame consisted of two roller assemblies joined by two frame elements
(Figure 19a). The rolle
r assemblies were allowed to pivot so that the robot can handle curved rail
sections. Acrylic was chosen as the material for the prototype because it was relatively inexpensive,
robust, and easy to manipulate. Locomotion was achieved using two rubber coate
d wheels driven by
hobby servos. The wheels must maintain contact with the rail at all times to provide a reliable means for
motion. The final frame design included a spring forced tilting mechanism which allowed the wheels to
follow the curvature of the r
Initial Frame Designs
The roller assemblies
(Figure 20) shown below were designed to minimize the overall width of the robot.
The hinges that were offset from the horizontal bracket allowed the robot frame sides(Figure 21) to be
the interior, thus making the roller brackets the widest part of the robot frame.
Initial Frame Designs
The frame sides were designed to minimize the length of the robot, while still accommodating the vital
components. The roller assemblies were attached to the ends of the frame side.
The servos were attached to the servo arms show
n below. The dimensions of the rectangular hole, 1.59”
X 0.78”, were the same as the cross section dimensions of many readily available servos. ¼” holes on the
sides were used to attach the servo arms to the frame sides.
functions that required consideration included vertical motion, position sensing and control,
power source, and rail switching. Vertical motion required high torque servos.
A simple incremental position sensing system was tested. This system consisted of
a reflective infrared
sensor and markers on the rail.
Conclusion and Future Work
Improvements and future work
A set up of rails and steel panels are
in Figure ____, so the final
magnetic robot can
displayed all together.
PICTURE COMING SOON
Display board of the MagneticBot and RailBot
Moore, T. (1985, Autumn). Robots for nuclear power plants.
, Retrieved from
World Nuclear Association, (2011).
Nuclear Power in the World T
Newmark, N., & Hall, W. (n.d.). Seismic Design Criteria for Nuclear Reactor Facilities, Retrieved from
Sharp, R., & Decreton, M. (1996). Radiation tolerance of components and materials in nuclear robotic applications.
AEA Technology: Elsevier Science Limited
. Retrieved from
Dodd, T., Atallah, K., & Lyne, I. (2010). Design and Optimization of Magnetic Wheel for Wall and Ceiling
, (August 4
Tache,, F., Fischer, W., Caprari, G., & Siegwart, R. (n.d.).
Magnebike: a magnetic wheeled robot with High Mobility for
Inspecting Complex Shaped Structures
. Unpublished manuscript, Autonomous Systems Lab, Zurich
Hirose, S., & Tsutsumitake, H. (1992). Disk Rover: A Wall
Climbing Robot using Permanent Magnet Disks .
Tache, F., F
ischer, W., Moser, R., Mondad, F., & Siegwart, R. (2007). Adapted Magnetic Wheel Unit for Compact
Robots Inspecting Complex Shaped Pipe Structures. (2007).
. Zurich, Switzerland.
We would also like to personaly thank our advi
sor Dr. Sean Brennan, as well as our Fellowship
Coordinator Melissa Marshall.
the The Pennsylvania State University
Department of Nuclear and Mechanical Engineering
Westinghouse, and Lowes