Abstract
This paper describes an application to support for designing LEGO robot arms in the 2-dimensional horizontal plane by using genetic algorithms. A LEGO robot arm consists of blocks, gears and motors, and the torque of each motor can be controlled by the RCX. The action rule and the structure of the LEGO robot arm are encoded as chromosomes of genetic algorithms, respectively. The fitness of an individual is calculated based on the trajectory of the hand of the generated robot through computer simulation. As a result of computer experiments, an individual which drew a target figure almost exactly was obtained. Then, we created a real LEGO robot arm to check the behavior of the best individual, and investigated the validity of the behavior observed in the simulation.
