Real World Hardware Evolution: A Mobile Platform for Sensor Evolution

Abstract

Although hardware evolution is becoming a more popular topic of research, the main focus of this research tends to be with recon configuring electronic circuits using evolutionary techniques. Taking a step back, my research looks at some of the problems of configuring autonomous, mobile systems for varying goals and environments. Concentrating on optical sensors, I am hoping to show that evolving the placement of sensors on the surface of the entity and the frequencies of light these sensors respond to will improve the entity’s performance in the environment.

This paper discusses some of the issues of working outside of simulation and presents a hardware platform I consider solves many of these to enable me to carry out the research described above. Unlike other mobile solutions, this platform was designed to be cheap, work in multiple entity environments and cope with large numbers of sensors (in my research, close to 200 sensors per entity) while still being extensible. This paper also covers issues such as processing power, environment management and entity interactions.