Today’s VIdeo Friday features Barbara Webb from the University of Edinburgh, Scotland, talks about using robots to model biological systems, including human and insect brains, and her cricket-robot experiments. This lecture was part of the ShanghAI Lectures. The ShanghAI Lectures are a videoconference-based lecture series on Embodied Intelligence. You can find the whole series from 2012 here.
The Insect Robot Group or Cricketlab is doing some fascinating work in this area.
Female crickets are able to locate mates by walking or flying towards the songs produced by males. We are modelling the neural circuits underlying this behaviour and testing the models on robots (including an outdoor robot). Funded by BBSRC.
Insects are capable of much more than simple reflex behaviour. We are investigating sensory integration and learning mechanisms in insect brains. Funded by EPSRC See Context Dependent and Multimodal Learning : From Insect Brains to Robot Controllers
Six-legged walking in robots does not yet have the flexibility of which insects are capable of. We have carried out behavioural experiments on stick insects and developed a dynamic simulation model of their control of turns. Funded by EPSRC See Neuromorphic Sensorimotor Integration for Legged Locomotion (NSILL)
Ants and bees are known to be able to locate an invisible nest site from the surrounding visual cues. Cockroaches and crickets can also use visual cues to locate a safe platform in a heated arena. We are developing behavioural paradigms and models of this capability. See Place memory in crickets
Insects have multiple sensory systems and a critical issue is how signals are integrated or interact in controlling behaviour. Work on this issue in our lab includes investigation of the interaction of phonotaxis and the optomotor response, and olfactory and visual responses in Drosophila
Methodological issues: Details here
Biomorphic wind sensing:
Crickets and cockroaches have wind sensitive hairs on two appendages on their abdomen. We are integrating MEMS hairs with analog VLSI neurons to create novel sensors for a robot able to replicate the escape behaviour of the insects. Funded by EPSRC.