Automatic for the People

Futurama's Bender, borrowed from

Futurama’s Bender, adapted from

In July 2014, the BBC, Guardian, FT and Wired covered a document that the Guardian described as “the UK’s first official robotics strategy”. But that makes it sound more official than it really is.

It’s been written – not by BIS or any other government department – but by members of the Robotics and Autonomous Systems Special Interest Group (RAS SIG) of the Technology Strategy Board (TSB).

There were 19 members of the RAS SIG who contributed: 3 academics (inc. one co-chair), 2 from EPSRC, 2 from Knowledge Transfer Networks – and 12 from the private sector (inc. the other co-chair).

Read this then, as a fairly apolitical piece of advice to government from industry.

The report suggests that AUVs could roam Loch Linnhe, and driverless cars through the roundabouts of Milton Keynes. That Sellafield and Boulby Mine would make great testing grounds.

It’s heavy on the skills, networks, partnerships and assets that could help the UK capture a bigger slice of the robot economy. It’s light to non-existent on what you might call ‘thought leadership’ of robots.

As automation becomes more prevalent, there are big implications for the workers of the future economy. Where technology once replaced brawn, future software might easily replace brain.

Don’t expect issues like these to be thoughtfully explored in the RAS SIG’s document. On page 8 the authors briefly acknowledge there could be pros and cons to robotics and autonomous tech:

Whilst [Robotics and Autonomous Systems are] a positive development, RAS technologies will raise concerns – some legitimate, some very far-fetched. Of course, films and novels have explored some of these issues, and there is potential for RAS technology to be used for both good and bad.

They also mention “There will be many reasonable and appropriate questions for public debate”. But ultimately, they see their role as campaigning for these technologies:

People often find it difficult to envisage the potential uses of robots until they are presented with specific examples, such as improved prosthetic hands, or on demand parcel delivery. Greater awareness of the multiplicity of roles RAS are capable of performing can therefore lead to an increase in support from the general public.

This is all fine, so long as it’s understood that this is the work of an industry group who are enthusiastic about the potential for the technologies they develop and want others to agree with them.

The net economic benefits to the UK’s economy might well be very significant. The Telegraph report the global market could be $70bn each year by 2025. Nothing wrong with that, either.

What is the problem then? It’s the lack of debate, understanding and preparation for the downside of automation. Stian Westlake writes in his introduction to NESTA’s recent pamphlet on robots:

It is hard for governments – or indeed for anyone – to accurately diagnose whether an innovation is sufficiently bad to ban. But politicians should encourage an open and informed debate about it, backed up with the capability to regulate effectively if necessary, and approach that researchers like Richard Owen and Jack Stilgoe have called ‘responsible innovation’.

The risks of automation aren’t unlikely or small, nor are the implications relevant only to the far future. On same day that the RAS SIG’s strategy was published, David Willetts was expected to say:

Robots have often been positioned as a thing of the future, but today’s strategy-launch emphasises the fact that they are very much of the here and now.

Technology moves oh, so quickly. Much faster than policy. Informed public debates and good public strategy on robots (and other technologies) matter. Can we have some, please?


Predicting the Future of Robotics with TRIZ


Image borrowed from Victor’s Stuff

TRIZ is a much less fashionable approach to innovation than, say, design thinking, but it does have a pleasing Russian moniker: теория решения изобретательских задач, or theory of inventive problem solving to English speakers.

TRIZ was developed by Genrich Altshuller, who analysed enormous numbers of patents, looking for repeating patterns in how engineers solved problems. From these, Altshuller developed several tools to help engineers solve problems more efficiently.

These include a ‘contradiction matrix’ and a set of ‘inventive principles’ which work together to prompt the brain into finding efficient ways to solve problems.

For example, something that must be both strong and lightweight is a contradiction: making things stronger often involves adding material. Entering these requirements into the contradiction matrix points to relevant principles, such as composite materials, or disposable parts.

Also from his study of patents, Altshuller developed eight trends that illustrate how many technologies evolve over time. One trend shows that while inventions often start as an immobile solid, the next generation is flexible, and the one after that is often a liquid or gas form – eventually the same problem may solved with a type of field.

To take a simple example (from Karen Gadd’s TRIZ for Engineers), blackboard pointers were originally rigid sticks, then developed joints to become hinged or telescopic devices, before (I think they skipped the liquid or gas phase) finally becoming electrical field devices in the shape of laser pointers.

Immobile System → Jointed → Many Joints → Fully Elastic → Liquid / Gas → Field

Does this trend help us predict the future of robotics?

In the late 30s, the Westinghouse Electric Corporation built an ungainly robot called Elektro, which was certainly pretty immobile (though it could smoke cigarettes). In ’54, a robot called the Unimate, which had the first jointed arm, was developed for General Motors. Modern industrial robots are now multi-jointed, and some androids appear completely flexible.

Perhaps with modular robots, such as MIT’s M-Blocks – which assemble themselves like jumping beans into the most appropriate shape for a specific task – we’re seeing the beginning of a more mature phase of robotics.

As modular robots develop and miniaturise, might we see nanoscale (a nanometre is 1,000,000mm) M-Blocks, so small they effectively form a robotic ‘field’ that can assemble itself into the most appropriate shape for the task at hand?