Barbara Webb and developing beetle-brained robots

[2020] Institute for Perception Action and Behaviour Director Barbara Webb to develop computational models for robots based on the navigation system of a dung beetle for her latest project, Ultimate compass.

Ultimate Compass robot
A model of the Ultimate Compass robot.

Barbara will be collaborating with colleagues at Lund University led by Marie Dacke on the project, entitled Ultimate Compass: unravelling the navigation system of a dung beetle. Barbara’s role on the project is to develop computer models of the dung beetle’s brain processes, which will be implemented on robots that can be tested in the natural habitat of the beetle. The team have received £151,139 for the project from Horizon 2020, an ERC funding programme for multi-national collaborative research and innovation projects.

The dung beetle is one of many navigating animals to use cues in the sky, such as the sun, polarisation or colour gradients, to find their way. There is also evidence to show that the beetles use night-sky cues like the moon and the Milky Way. As such, dung beetles provide the opportunity to analyse the ability of navigation using sky cues at multiple levels: ecology, sensory physics, neural processing and motor control. The project explores how to integrate these cues of varying reliability in order to make good behavioural decisions, a key issue for cognition and artificial intelligence.

Insect robotics review article published in Science magazine

The Ultimate compass project is not Barbara’s first work on beetle-brained robots: she recently wrote a perspective article ‘Robots with insect brains’ which has been published in Science Magazine, one of the world’s top academic journals. The article reviews and summarises the current state of research into insect robotics, highlighting the ability of computer models to copy the processes of the insect brain and be implemented on robots, which enables testing of real bodies interacting with real environments.

Science magazine is a peer-reviewed academic journal of the American Association for the Advancement of Science. It has a current subscriber reach of 130,000, however institutional subscriptions and online access serve a larger audience giving a much higher estimated readership of 570,400.

The ability to build a machine that replicates insect’s brains functions would be the ultimate test of mechanistic understanding of how it works. Prominent examples in the area of insect robotics include the visual target tracking of dragonflies being replicated on a (wheeled) robot platform performing active pursuit or insect target tracking behaviour examined in the praying-mantis–inspired “mantisbot”.

Future queries may include further insight into the mushroom body (MB), the region of the insect brain known to be involved in associative learning of the value of olfactory stimuli. Multiple modelling studies and some robot applications have shown that it can support pattern learning by encoding inputs as sparse activation of a small subset of a larger neural population and correlating with a reward signal.

Modelling the whole insect brain is still on the table: several groups, inspired by detailed D. melanogaster brain wiring diagrams, are now pursuing this target. More detailed brain models however are only likely to lead to insights if they are grounded in understanding behaviour.

Related links

'Robots with insect brains' perspective article in Science

Barbara Webb's personal page