22-02-2013, 03:28 PM
Using Robots and Contract Learning to Teach Cyber-Physical Systems to Undergraduates
[attachment=51488]
Abstract
Cyber-physical systems are a genre of networked
real-time systems that monitor and control the physical world. Examples
include unmanned aerial vehicles and industrial robotics.
The experts who develop these complex systems are retiring much
faster than universities are graduating engineering majors. As
a result, it is important for undergraduates to gain exposure to
these kinds of complex systems. This paper describes UPBOT, a
robotics testbed hosted at the University of Portland, Portland,
OR. The testbed features an extensible robot built from an iRobot
Create chassis and a computationally powerful embedded system
equipped with a wireless card. In the Spring 2012 semester, the
testbed was used for a course project designed and assessed in
the style of contract learning. Overall, students were enthusiastic
about the hands-on and self-paced nature of the course, but some
were concerned by the level of freedom. The lessons learned may
inform other educators in project design in robotics education.
INTRODUCTION
CYBER-PHYSICAL systems are distributed networks
of real-time systems that monitor and control the
physical world. They include automobiles, avionics, and industrial
robotics. Over time, these systems have increased
in complexity, evolving into larger systems of systems. This
complexity is compounded by an aging work force whose
retirement rate is outpacing the rate of US students majoring
in the engineering fields [1], [2]. The people who understand
these very complex systems are retiring from companies such
as Ford and Lockheed Martin, and from the military, while
students are turning away from engineering degrees [3]. As a
result, educators must work harder than ever. Students require
some kind of hands-on exposure to systems that are complex
enough to be relevant, interesting enough to be motivating, and
accessible enough to be engaging.
MOTIVATION
To illustrate the need for a learning platform for cyber-physical
systems, consider three concrete examples.
Example 1: The cockpit of an American Predator or Reaper
drone is a small office room wallpapered with 14 computer
screens. Pilots in the Creech Air Force Base in Nevada use
these displays to control unmanned aerial vehicles in Pakistan,
Afghanistan, Yemen, and Iraq. These drones are used to deploy
missiles, search for disaster survivors, or provide surveillance.
In October 2011, despite the “air gaps” between the classified
drone networks and the Internet, a keylogging virus was reported
on the computers charged with operating the drones [6].
Example 2: Modern automobiles make use of multiple microcontrollers
that control braking, emissions, lighting, and infotainment.
These microcontrollers communicate over multiple
controller area networks (CANs). Documented in [7] are multiple
approaches for an attacker to remotely compromise the automobile
using Bluetooth, a CD player, or the telematics unit
(e.g., OnStar). The authors demonstrate these attacks on their
own car and propose a scenario in which a clever thief could
compromise numerous cars.
CONTRACT LEARNING
To motivate students, engineering educators must work hard
to design course activities that are fun, hands-on, and relevant.
To achieve these motivational objectives, it is not enough to
assign problem sets from the course textbook. However, just
designing fun activities can be a mistake. As described by
Wiggins and Tighe in [10], when thinking only like an activity
designer, an educator might mistakenly ask, “What would be
fun and interesting?” Instead, engineering educators must also
think like assessors and ask, “Against what criteria will we
appropriately consider work and assess levels of quality?”
Hence, the challenges of designing a hands-on course project
are twofold: motivation and assessment.
Results
The University of Portland’s small class size offers opportunities
to develop intensive hands-on activities, as described
above. It also means that it is difficult to gather statistically significant
data for a technical elective course offered, at best, every
two years and taken by only a handful of students. While the author
acknowledges the anecdotal nature of the results, this section
briefly summarizes the quantitative and qualitative results
from the student evaluations, exempted under the University of
Portland Institutional Review Board.
At the end of the semester, the students were given the standard
online evaluation offered to all students on campus. This
evaluation is optional, and students complete the evaluation outside
of class during a two-week period at the end of the semester.
The evaluation contains five quantitative questions to which
students reply using one of five discrete options ranging from
“Strongly Agree” to “Strongly Disagree.” The online evaluation
also contains two open-ended questions. Sixteen of the 18
EE439 students completed the online evaluation.
RELATED WORK
There is little denying that robots are fun. Many college
courses [17], [18], including the University of Portland’s own
first-year EGR110 Introduction to Engineering course [19],
use LEGO Mindstorms as a platform for entry-level engineering
education. LEGO has valuable strengths; robots are
easy to assemble and are programmable using a graphical,
LabView-like interface. However, there are computational
limitations to LEGO that make it difficult to use as a platform
for nonintroductory courses.
Around the country, other platforms are available for a
variety of student mastery levels. Inspiration for the robot
platform used in UPBOT came, in part, from [20], which describes
the robotics field for the young novice. Its open-source
workbook features an iRobot Create-based platform that uses
a variety of options, including an embedded system using
the Player robot device interface library or a BAM Bluetooth
device interacting with the Microsoft Robotics Studio.
CONCLUSION
Cyber-physical systems are a genre of networked real-time
systems that monitor and control the physical world. It can be a
challenge to develop relevant course projects that expose undergraduates
to the complexities of such systems while simultaneously
motivating them to really grapple with the technical realities
presented by such complexity. The combination of a wellcrafted
testbed, an extensible robotics platform, and a contract
learning-based course project is one way to meet this challenge.
[attachment=51488]
Abstract
Cyber-physical systems are a genre of networked
real-time systems that monitor and control the physical world. Examples
include unmanned aerial vehicles and industrial robotics.
The experts who develop these complex systems are retiring much
faster than universities are graduating engineering majors. As
a result, it is important for undergraduates to gain exposure to
these kinds of complex systems. This paper describes UPBOT, a
robotics testbed hosted at the University of Portland, Portland,
OR. The testbed features an extensible robot built from an iRobot
Create chassis and a computationally powerful embedded system
equipped with a wireless card. In the Spring 2012 semester, the
testbed was used for a course project designed and assessed in
the style of contract learning. Overall, students were enthusiastic
about the hands-on and self-paced nature of the course, but some
were concerned by the level of freedom. The lessons learned may
inform other educators in project design in robotics education.
INTRODUCTION
CYBER-PHYSICAL systems are distributed networks
of real-time systems that monitor and control the
physical world. They include automobiles, avionics, and industrial
robotics. Over time, these systems have increased
in complexity, evolving into larger systems of systems. This
complexity is compounded by an aging work force whose
retirement rate is outpacing the rate of US students majoring
in the engineering fields [1], [2]. The people who understand
these very complex systems are retiring from companies such
as Ford and Lockheed Martin, and from the military, while
students are turning away from engineering degrees [3]. As a
result, educators must work harder than ever. Students require
some kind of hands-on exposure to systems that are complex
enough to be relevant, interesting enough to be motivating, and
accessible enough to be engaging.
MOTIVATION
To illustrate the need for a learning platform for cyber-physical
systems, consider three concrete examples.
Example 1: The cockpit of an American Predator or Reaper
drone is a small office room wallpapered with 14 computer
screens. Pilots in the Creech Air Force Base in Nevada use
these displays to control unmanned aerial vehicles in Pakistan,
Afghanistan, Yemen, and Iraq. These drones are used to deploy
missiles, search for disaster survivors, or provide surveillance.
In October 2011, despite the “air gaps” between the classified
drone networks and the Internet, a keylogging virus was reported
on the computers charged with operating the drones [6].
Example 2: Modern automobiles make use of multiple microcontrollers
that control braking, emissions, lighting, and infotainment.
These microcontrollers communicate over multiple
controller area networks (CANs). Documented in [7] are multiple
approaches for an attacker to remotely compromise the automobile
using Bluetooth, a CD player, or the telematics unit
(e.g., OnStar). The authors demonstrate these attacks on their
own car and propose a scenario in which a clever thief could
compromise numerous cars.
CONTRACT LEARNING
To motivate students, engineering educators must work hard
to design course activities that are fun, hands-on, and relevant.
To achieve these motivational objectives, it is not enough to
assign problem sets from the course textbook. However, just
designing fun activities can be a mistake. As described by
Wiggins and Tighe in [10], when thinking only like an activity
designer, an educator might mistakenly ask, “What would be
fun and interesting?” Instead, engineering educators must also
think like assessors and ask, “Against what criteria will we
appropriately consider work and assess levels of quality?”
Hence, the challenges of designing a hands-on course project
are twofold: motivation and assessment.
Results
The University of Portland’s small class size offers opportunities
to develop intensive hands-on activities, as described
above. It also means that it is difficult to gather statistically significant
data for a technical elective course offered, at best, every
two years and taken by only a handful of students. While the author
acknowledges the anecdotal nature of the results, this section
briefly summarizes the quantitative and qualitative results
from the student evaluations, exempted under the University of
Portland Institutional Review Board.
At the end of the semester, the students were given the standard
online evaluation offered to all students on campus. This
evaluation is optional, and students complete the evaluation outside
of class during a two-week period at the end of the semester.
The evaluation contains five quantitative questions to which
students reply using one of five discrete options ranging from
“Strongly Agree” to “Strongly Disagree.” The online evaluation
also contains two open-ended questions. Sixteen of the 18
EE439 students completed the online evaluation.
RELATED WORK
There is little denying that robots are fun. Many college
courses [17], [18], including the University of Portland’s own
first-year EGR110 Introduction to Engineering course [19],
use LEGO Mindstorms as a platform for entry-level engineering
education. LEGO has valuable strengths; robots are
easy to assemble and are programmable using a graphical,
LabView-like interface. However, there are computational
limitations to LEGO that make it difficult to use as a platform
for nonintroductory courses.
Around the country, other platforms are available for a
variety of student mastery levels. Inspiration for the robot
platform used in UPBOT came, in part, from [20], which describes
the robotics field for the young novice. Its open-source
workbook features an iRobot Create-based platform that uses
a variety of options, including an embedded system using
the Player robot device interface library or a BAM Bluetooth
device interacting with the Microsoft Robotics Studio.
CONCLUSION
Cyber-physical systems are a genre of networked real-time
systems that monitor and control the physical world. It can be a
challenge to develop relevant course projects that expose undergraduates
to the complexities of such systems while simultaneously
motivating them to really grapple with the technical realities
presented by such complexity. The combination of a wellcrafted
testbed, an extensible robotics platform, and a contract
learning-based course project is one way to meet this challenge.