SOCIAL -Collaborative agent development

In this paper, we outline the development of a tangible Collaborative Autonomous Agent (CAA), capable of locating and repairing scale formations in tanks and pipes within inaccessible environments. These agents are capable of perceiving and exploring their environment using onboardsensors and actuators; they also have the added ability to communicate with other agents underwater, and can learn their behaviour at both an individual and social level. Each agent is based on the biologically inspired Spiking Neural Networks (SNNs) that communicate with either pulses or spikes. The SNNs are evolved using Genetic Algorithms (GAs). These SNNs have been successfully embedded on FPGA hardware within the agent, which is ideal for intensive computational purposes. Due to the nature of the agents being developed, a highly modular electronics platform incorporating FPGA technology, an ISM band wireless communications platform, sensors conditioning circuitry and motor actuation capability has been developed in a miniaturized form factor.