Comment & Analysis
In computer science there have been interesting models made of biological solutions over the last 20-30 years, such as neural networks, cellular automaton, and evolutionary computing. At Durham University we decided to build on this research to develop ‘electronic stem cells’, which can be built into electronics to repair and restore functionality to damaged hardware.
Based on algorithms called cellular automaton, these are able to mimic the specialised stem cells that can repair any tissue in the human body. We are now finding ways to build a healing capacity in electronic systems. Healing is when a system can, within its own resources and timescale, affect a repair that allows it to continue functioning as before.
We are attempting to recreate the kind of self-healing function that occurs when you break your arm or cut yourself. The body has resources to effect a ‘repair on the road’ without going into hospital.
Importantly, this repair process is not data correction. We are moving beyond taking the information that flows through a computer system and correcting it. It is about physically repairing the infrastructure of the electrical calculation.
Our electronic stem cells, or cellular automaton algorithms, seem to be good at restoring expected functionality in an electronic device. A cellular automaton is a collection of ‘coloured’ cells on a grid of specified shape that evolves through discrete time steps according to rules based on the states of its neighbouring cells. The rules are then repeated for as many time steps as desired.
So in a computer program you could produce an image of the Union Jack flag with its colours. With a few cosmic rays directed at the computer you could destroy that picture and change the values of some of the red and blue pixels.
If that computer were protected by electronic stem cells it would realise there was an error and try as hard as it could to get back to the image it should be storing. From there it is a small step to get to a computer that will, when damaged, continue to fight to get back to its full state of health.
We have recently been running tests on flight deck avionic systems where an error in a computer could be catastrophic. At this stage the tests are on simulations where we run flight failures with and without the self-repairing structures. With the electronic stem cells it is a much less drastic state of affairs. Most of the time the electronics can recover the aeroplane and keep it, if not fully functioning, then at least well enough to limp home.
The aerospace sector is a promising field for further trials, along with critical infrastructure such as nuclear power, and wind turbines. I imagine the first time we will see this working in the real world is in planes where the self-repairing system would act like a second advisory autopilot, reporting errors and recommendations to the autopilot or pilot. Much further in the future the whole aeroplane’s safety could be handed over to the supervisory autopilot.
Our major aim is to ensure that electronic products last longer. This is apt with the rise of autonomous car technology. We could potentially have a Google car that is fully maintenance free.