York Robotics Laboratory
Towards the Autonomous Evolution of Robotic Ecosystems
Emma Hart delivered a keynote on ARE during IEEE CEC 2019.
Abstract
From its very beginnings, Evolutionary Computation has been used as a tool to evolve artefacts, starting with the very first optimisation of a joint plate at the Technical University of Berlin in 1965, quickly followed by evolution of the often-cited two-phase nozzle in1968. Since then, advances in computing (CPU power, simulation engines), materials science, and engineering (3D printing, automated assembly) have considerably enhanced our ability to evolve artefacts: these range from design of functional objects such as satellite antennae, through creative design of chairs and tables, to design of active artefacts such as robots that dynamically sense and interact with their surroundings.
In this talk, I will give a brief history of the evolution of artefacts, leading to evolutionary robotics, and finally to a recent collaborative project that aims to deliver a disruptive robotic technology in which robots are created, reproduce and evolve in real-time and real space. The long-term vision is of a technology that enables the evolution of entire autonomous robotic ecosystems that live and work for long periods in challenging and dynamic environments without the need for direct human oversight, e.g. in outer-space, or decommissioning a nuclear reactor. Rather than being designed and manufactured, the morphologies and control processes of robots will emerge as a result of evolutionary processes, continuously changing both their form and behaviour over-time.