Pint of Science: Going Legless with Arachnidbots and Neural Networks | Techilink

Pint of Science is the brainchild of two doctors who, in 2012, organized an event called “Meet the Researchers” by inviting patients with motor neuron disease to their laboratories to raise awareness of the research that was taking place at the time. It was a natural progression to bring cutting-edge research – from some of the UK’s top scientific institutions – to the people in a much more accessible environment: the pub.

In 2013, Michael Motskin and Praveen Paul held the first Pint of Science festival, and in the years that followed, the idea caught on around the world. It is primarily run by volunteers and a community of graduate and postdoctoral researchers who are boldly willing to share their findings with perhaps a bunch of drunks.

Once a year, the festival takes place simultaneously in pubs around the world for three days in the month of May. There are usually two talks per evening, each lasting up to 30 minutes, with time for questions and answers afterwards and even a half-time quiz if you’re lucky.

It could just be my growing and probably unreasonable automatonophobia, but when I heard the news that the Science Museum is trying to raise money to reassemble Eric, the first humanoid robot originally built in 1928, it seemed only right to learn more about robot articulation and locomotion.

Self-contained adaptive robots? Incy wincy spiders lead the way

On my first visit, I found myself with a pint of cider firmly in hand at the Rugby Tavern in Farringdon, London. Luckily I had just finished my cheesy chips when one of the organizers told the drinkers it was time for a bite to eat. Spinning a pint… with robots presented by Michelle Reeve (PhD student at the Royal Veterinary College Structure & Motion Laboratory). Reeve has been studying the movement and gait of spiders to find a way to improve the mobility and flexibility of the next generation of robots.

She introduced the public to her research beyond the working robots we have today, like the multi-colored Asimo, which can climb and pick up objects, but probably isn’t great in a crisis situation just yet. Or the terrifying Atlas, which tries to approximate human movement with incredible balance and flexibility. Reeve’s work instead focuses on robots with a much more beastly set of protocols: think Boston Dynamics’ robots, such as the incredibly fast Wildcat, heavy cargo carrier Big Dog, or cockroach on tank tracks—like REX used to navigate rough terrain.

Reeve also introduced robots made from polymers based on worms and squid that can squeeze through the tiniest gap. But robots aren’t perfect: they often struggle with doors and can get damaged, which can defeat their purpose. If a search and rescue robot deployed to access environments unsuitable for humans needs help from a human after hitting a snag and breaking, for example, it quickly becomes pointless.

Animals, on the other hand, seem capable of sorting themselves out in such predicaments. There are many examples of lizards that regenerate their tails and spiders that can regrow legs. Animals have evolved a variety of gaits that are basically footfall patterns, such as trotting, galloping, and stalking, which help reduce energy expenditure or navigate different terrains. Animals adapt these gaits when injured, which is where Reeve’s spiders come in.

Spiders live in a variety of terrains and across many species have adapted to a variety of gaits and leg movements, such as jumping and building webs, and they adapt to the loss of legs. Spiders that amputate an appendage as a result of an injury, which is called an autotomy, change their gait to an alternating tripod to stabilize. The same mechanism could be directly applicable to robotics.

Reeve used a commercially available T8X, which can be programmed to allow researchers to alternate robot gaits by applying the spider data to understand the adaptability of those eight-legged arachnids. Reeve is not alone in thinking that this research could lead to more independent and flexible robots. She highlighted relevant research published last year in the Nature diary, which involves an adaptive six-legged robot that works through trial and error and chooses the mode of movement that hurts it the least.

For example, independent adaptive robots could patrol large areas of forest looking for wildfires without relying on human attention for repairs, Reeve concluded. Inspired by this talk it was a quick decision at half time to name our quiz team Spiderpig.