Hardware Evolution System AdAM

The Adaptive Architecture Methodology (AdAM) [3] is a platform utilized in a hardware evolution experiment. We constructed the hardware evolution system AdAM based on Hardware Description Language (HDL). This system automatically generates HDL and can evolve HDL—we briefly explain the operation of the system here.

The population of chromosomes is generated according to the HDL production rule. Next, the fitness of each individual is evaluated by the operation simulation at the HDL level. The individual that remains in the next generation is selected according to the fitness evaluation. A genetic operation is given to the chromosome of the selected individual, and the next generation’s chromosome population is made. Such evolutionary processes eventually result in individuals that have the desired hardware behaviors. Various experiments are done based on this system. We implemented a diploid chromosome and a progressive evolution model based on this system—the outline of the diploid chromosome and the progressive evolution model are explained here.

Diploid chromosome. The chromosome of the basic AdAM system is a haploid chromosome. We extended the chromosome of the basic system to a diploid chromosome [1]. The extension was necessary to give the system enough genetic diversity to evolve under environments that change with time.

Most multicellular organisms have diploidy, and it is thought that this maintains population diversity and improves the survival of species in a fluctuating environment. Since the dominant-recessive heredity shown by an organism with diploid chromosomes is important, we extended the chromosome of the basic system to a diploid chromosome for a system mechanism that maintains diversity. Genetic operations were also extended, and meiosis and fertilization were added.

Progressive evolution model. The basic idea of the progressive evolution model [2] is that organisms evolve while acquiring new functions to match environmental changes. We therefore believe that environmental changes drive evolution, so our idea actively uses environmental changes to accelerate evolution. In the progressive evolution model, evolution occurs in environments that change in a stepwise manner toward the final target environment (TE).

The purpose of the model is to divide a large “hurdle” into a series of small steps so that the evolutionary process can easily handle the hurdle. We have extended the operation simulation of the basic system. The environment has been progressively changed from an easy one to a more difficult one. We call the intermediate easy environments Progressive Environments (PE). This model is explained as an example of an artificial ant problem as shown in Figure 1.

The black square in Figure 1 shows the food of an artificial ant. In our problem, artificial ants adapt to a certain environment so that they can quickly and effectively gather food. The total time required for the ant evolution in a progressively changing environment could be shortened more than the ant evolution in the target environment.

We implemented the diploid chromosome and the progressive evolution model based on the AdAM system. In the future, we will consider the construction of a system that replaces the operation simulation of the basic AdAM system with actual hardware.

Back to Top

Back to Top

Back to Top

Back to top