Abstract

Bridging the reality-gap and designing robots that can work in natural environments is a critical challenge. Over the past decade we have taken inspiration from eusocial insects to develop robot teams that can collaboratively solve foraging problems in the real world. This talk covers some theoretical results and examples of field deployment in the context of extreme environments where robots are not only beneficial but required to make progress: 1) The Moon and Mars colonisation initiatives, 2) Measuring and collecting carbon-dioxide emissions from actively erupting volcanos, and 3) Collaborative transport of radioactive containers. I will also present our recent work on scalability of the mammalian immune system that results from extreme first-passage analysis for biologically realistic numbers of cells. The interest of Computer Scientists and Roboticists in biological movement patterns is in how we can co-opt them for our own needs and design efficient algorithms, but as robots become more sophisticated and interact with complex nature environments they become more difficult to predict and the line between biology and bioinspired blurs. Conversely, there appear to be examples where apparently complex biological phenomina can be approached with realtively simple theory.