Researchers in Sheffield and Bristol have built a robot that uses rat-like whiskers and a rat brain model to explore and map its environment.
Rather than using the whiskers as simple collision probes, the whiskers are moved rapidly ('whisked') to scan the environment, with this movement modified by the surfaces and objects encountered to extract more detail.
"They are moving back and forth at 3Hz, sweeping through an arc of 60 degrees," Professor Tony Prescott from the University of Sheffield's department of psychology told Electronics Weekly. "Rats sweep at 5-20Hz through the same angle."
See also: The UK Electronics Research Map
The University of Sheffield is working with the Bristol Robotics Lab - a partnership between the University of the West of England and the University of Bristol.
Sheffield is working on the brain model and control software, with Bristol developing mechanics, electronics, and a hardware neural network.
The outcome is an increasingly sophisticated physical and computer model called ScratchBot, which is a contribution to the pan-European 'ICEA' project which aims to develop biologically-inspired artificial intelligence systems.
At the moment, ScratchBot's intelligent is partly on-board and partly post-activity off-line processing.
"We have got all the various bits for texture and distance-to-object. We are starting to do curvature and shape, and we are working on a model of the hippocampus that will store two-dimensional information."
Prescott's team works to understand natural systems by building models based on known science, then modifying the models until they behave like the natural system.
This isolates a possible mechanism for the natural system, and produces algorithms that can be used to perform practical tasks.
"We tend to build biomimetic systems, then design a reduced version with more numerical algorithms that does the same thing," Prescott explained.
ScratchBot has nine sensing whiskers on each side (rats have 30). Each has a Hall-effect movement-vibration sensor on its base, and they can be swept back and forth in groups of three, by three electric motors on each side.
This sweeping, combined with three degrees of freedom in its neck - pitch, yaw, and elevation - give ScratchBot all the movement it needs to scan all surfaces within reach.
All real-time processing through on-board hardware is planned.
There are many nocturnal, underground, and underwater animals for whom whisking is more important that sight, pointed out Prescott. And eventually, it is hoped that techniques from the research will allow robots to move easily through dusty and dark environments like collapsed buildings.
Can the brain-whisker model with its abstract data be interfaced with conventional computers?
Yes, said Prescott: "The current robot is already a mixture of biomimetic and conventional numeric algorithm computing."
ScratchBot builds on previous whisker research at Sheffield.