IEEE Robotics & Automation Magazine - September 2017 - 163

students. They noted that, at one end of the spectrum, there
are multiple popular books for hobbyists and middle
schoolers and, at the other end, there are several books
appropriate for advanced Ph.D. students due to the level of
mathematical maturity required. Several participants bristled at the idea of using the twist-wrench concept or group
theory in an introductory course. Quite a few faculty members write supplementary notes to bring students from
other disciplines up to speed on background topics that are
needed to tackle the text.
The second difficulty mentioned was finding a single text
that covers all of the topics that a given instructor deems to be
important. Considering Figure 2, it appears that many
instructors focus on two or more subdisciplines of robotics in
their course that are unlikely to be covered in a single book,
e.g., feedback control and motion planning. They combine
multiple texts and self-authored notes to add material that
they believe should be covered. This frustration is no doubt
tied to the issues discussed in the "Disciplinary Affiliations
and Backgrounds" section regarding the diversity of student
backgrounds and the breadth of the discipline. The final
group of comments suggests that 1) instructors are seeking
books that include problem sets and practical exercises that
can be used to complement the lecture component of the
course and 2) a great many books are deficient in this regard.
When asked "Do you use any free online courseware?," only
10.8% answered affirmatively. Given the dissatisfaction with
traditional texts and the recent rise in the popularity of massive open online courses (MOOCs), this is surprising. The
online resources that were mentioned included Introduction
to Robotics and Robot Vision by Peter Corke; Control of Mobile
Robots by Magnus Egerstedt; and a free text, Planning Algorithms by Steven LaValle. The website www.roboticscourseware.org collects notes, presentations, and assignments from
multiple universities in a single place.

which is interesting in light of the fact that there was no
official MATLAB robotics toolbox until Release 2015A
(although a third-party toolbox dates back to 1996).
C and its variants are cited as the second most popular
option (52.3%).
Free open-source software libraries created specifically for
robotics applications such as ROS (27.7%), OpenCV (16.9%),
and Point Cloud Library (3.1%) still considerably lag behind
these more traditional languages. In our opinion, this has to
do with the challenges that these tools present for the novice
programmer. The Linux operating system, object-oriented
programming concepts (C++, in particular), and linking to a
large set of libraries within an integrated development environment are all, essentially, prerequisites to learning the particulars of ROS, for example. Since 26% of the students are
mechanical engineers, they are unlikely to have had any formal exposure to these skills. In contrast, MATLAB is frequently taught to mechanical and electrical engineers and is
easy for a computer science student to learn.
The variety of responses to the question "What hardware
platforms do you use?" is notable. A total of 34 different
commercially available platforms were cited, and six faculty
members indicated that they use custom-made platforms.
The only platforms mentioned by more than two respondents were Lego Mindstorms (eight); iRobot Create (seven);
the open hardware Turtlebot (four); Adept Pioneer (four);
and various Lynxmotion kits (three), some of which are
shown in Figure 4. These are all either mobile robots or
desktop kits, and most sell for under US$1,000. A recurring
theme in the comments section is the need for lower-cost,
small-form-factor, high-quality articulated manipulators.
Future Challenges
The final question asks "What will robotics education look
like in ten years?" There were as many opinions as there were
participants, with little agreement among them. Answers
ranged from "just like it does now" to more ambitious visions
that include a heavy reliance on smart phones as computing
platforms; a shift toward small, affordable humanoid robots;
or primarily MOOC-based programs. I am confident that

Hardware and Software Platforms
A large majority of the instructors (81.5%) indicate that their
course includes a hardware-based laboratory component;
73.8% also include an open-ended final project. Clearly,
roboticists consider implementation to be an important component of the educational
experience. Unfortunately, the
third most common response
(21%) to the question "What are
the biggest challenges in robotics education?" was a lack of
departmental funding and/or
space to provide each student
group with a dedicated lab setup.
When asked for the computing languages and tools that
they use in their course, teachers
cited the most popular choice Figure 4. In terms of laboratory hardware, a few small-scale mobile platforms dominate the
to be MATLAB (61.55%) by far, market. (Photo courtesy of Jenelle Armstrong Peipmeier and Joel Esposito.)
SEPTEMBER 2017

IEEE ROBOTICS & AUTOMATION MAGAZINE

163

http://www.roboticscourse http://www.ware.org

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - September 2017