Building Robots: Expectations and Experiences

In tlie four years that llie MU Mobile lbbot Project has been 1 General Theme at the Start in existence, we have built ten robots that focus research in various areas, aimed primarily towards discerning what is involved in building intelligent, useful autonomous creatures. Many of the preconceived notions entertained before we started turned out to be misguided, many issues we thought would be hard have worked successfully from day one, and subsystems we imagined to be trivial have become tremendous time sinks. Oddly enough, one of our biggest failures has led to many of our favorite successes. The general problem we set out to solve four and a half years ago was how to build a brain, or, to answer the question of what it would take to build something that we would consider clever. What were the essential components that would be needed to create an intelligent entity and how should those components be put together? The ideas we started with took a route that was different from the traditional thinking in Artificial Intelligence at that time. Namely, our From early meter-high, offboard-computer based robots, to sleek walking creatures, soda can collection machines and one cubic inch bugs, the MIT Mobot Line exhibits a wide variety of talents, sensing strategies and locomotion approaches. Central throughout is a common methodology for organizing the sensors, actuators and computational elements to effectively control complexity. A basic tenet is that it is important to build complete systems that exist in real world noise. This avoids the trap of building