Robot platforms have played a fundamental role in the field of artificial intelligence (AI) for more than 30 years. Yet it is only recently that physically embodied agents have become a viable tool in the undergraduate AI classroom. Examples of the flurry of activity in this area include competitions and exhibitions, the growing options for low-cost robot hardware and software, and a number of recent workshops and symposia. This special issue of AI Magazine grew out of the 2004 AAAI spring symposium on Accessible, Hands-on AI and Robotics Education. In this article, we seek to showcase how robots have influenced both the curriculum and practice of teaching AI at the undergraduate level.
This survey article first overviews a number of robot platforms and presents trade-offs in choosing among them. We then highlight the variety of AI curricula supported by low-cost robotic platforms. We conclude with a summary of the engaging and active communities that support robotic competitions and exhibitions. These robot-based components, curricula, and communities, we hope, broaden the resources available to educators, as we all invite students to share our enthusiasm for AI.
Robot Platforms for AI Education
Stuart Russell and Peter Norvig frame their widely used AI text through a paradigm of intelligent agents (Russell and Norvig 2003). Such an approach resonates with students, all of whom have deep experience with (and as) intelligent agents. Yet nearly all of that experience is with embodied intelligent agents, and this familiarity makes robots a strong motivator of AI. This embodiment contrasts with the majority of computer science subfields, in which computers interact with the physical world very differently than we do. What's more, for AI educators, robotic hardware is not only a hook that can draw students to the field, but a fundamental facet of the AI endeavor. The challenge is to find a set of hardware and software resources that serve both as motivation and as tools to advance, not limit, the AI that students pursue in an academic course of study.
Today there exists a large and growing selection of robotic platforms suitable for conveying and investigating fundamental AI topics. Figures 1, 2, and 3 summarize some of these resources and their capabilities, with particular attention given to newer models and those widely employed at the undergraduate level.
These sensors are available from a number of retailers including HiTechnic Products and Mindsensors.com for the RCX-compatible sensors and Acroname for those sensors not specifically tailored to the RCX's Lego interface.
Before discussing the platforms listed in figure 1, it is worth mentioning a family of low-cost robotics resources we have omitted: those dedicated to teaching the electrical and mechanical engineering that underlies most contemporary robotics, such as the basic stamp microcontroller (Kuhnel and Zahnert 1997). Undergraduate AI does not ignore the impact of such design decisions but instead focuses on the computational challenges those decisions create. In the context of AI education the hardware/software interface, that is, the ease with which students can interact computationally with a robot and investigate how their algorithms behave, is a crucial criterion for evaluating robotic platforms.
Hardware and Software
A key advantage of the two most popular platforms, Lego Mindstorms (or RCX brick) and the Handy Board, is the variety of ways in which students can program them. C-like languages and Java subsets are available for the Mindstorms through the BricxCC and LeJOS firmware upgrades. Both are open-source projects with substantial deployment. Interactive C is the default computational interface on the Handy Board. A commercial Java implementation, RoboJDE, is available for the Handy Board from RidgeSoft, LLC. (1) These two platforms' large user communities breed support for a wide variety of interfaces: of particular note is the Lisp interface to the Lego RCX brick described in detail later in this AI Magazine issue. …