Learning embedded and real-time systems via low-cost mobile robotics

The importance and impact of embedded and real-time computing systems on today's society far exceed that of traditional stand-alone computers; it is hard to think of new device or system that does not embed some combination of silicon-based intelligence, sensing, and communication. A parallel trend is the growth of high-level, abstract design of these systems and associated languages and CAD tools, driven by increasing processor speed, time-to-market requirements and the complexity of applications. However, many courses in embedded systems focus on low-level issues such as addressing, interrupts and interfacing. In this paper, we describe a new direction: a course with the goal of motivating students to learn the abstract concepts that underly the design of these systems via experiments that require the interaction of robots with the physical world. In this studio-format course, most conventional learning takes place outside the class, while small student teams design, build, and evaluate autonomous mobile robots in the classroom/laboratory. To keep costs down, mobile robots are created using LEGO parts and programmed in the high-level NQC language using the Robolab RCX microcontroller module. As the semester proceeds, students tackle an array of interrelated problems that motivate the study of sensor signal processing, control, scheduling, and resource sharing. In a final project, the students tackle a distributed intelligence project in which an odometry-equipped robot communicates with a PC-based program that tracks the robot's position. To encourage adoption by other electrical engineering and computer engineering programs, a detailed description of the required resources and their cost is included.