Building a basically functioning robot in EECS192 typically takes 5 weeks of the course. Students use the remainder of the course to improve sensors, system integration, and algorithms. Debugging the whole system of course takes time, and the more complicated the system is, the longer the debugging takes. Students work in teams of 2 or 3 students, to divide the work. Experience shows that simple designs take less time to build, and work better!
The design process is broken down into manageable steps through design checkpoints. Each design step is preceded by a lecture covering the main ideas and principles. Here is an outline of the design checkpoints for the first weeks:
Thus after 5 weeks, the vehicle hardware is mostly done. For Spring 2001, we will mostly be using the computer and cars already built from previous semesters. The remaining hardware to add is a line sensor so the robot can stay on the race track. Algorithm and control strategy development take up the next 9 weeks. Again, there are well-defined design check points to ensure timely progress and keep the project scope bounded.
Each robot race car will be individually timed as it follows a line or buried wire laid out on a 100 meter path in a large arena. The path is not known until the time of the race, and has many curves and self-crossings. Every team is using the same motor and batteries, so competitive advantage comes from using smarter algorithms that are better at keeping the car on the race track. A time penalty is used for vehicles that stray too far from the line and knock over traffic cones, so simply using maximum acceleration will not be a good strategy.
R. Fearing