The scientific briefing with Luke Boulter
They can walk, talk and carry a tray of coffee: the new generation of robotics is here, but how far do they live up to 1970’s sci-fi predictions: are robots set to take over the world, or are we taking giant steps into mechanical servitude?
It is tricky at best to exist nowadays without having experienced robotics at some point, from the car you may drive, constructed on a robotic production line, to the antibiotics you take, cultured using a robotic system. It will, however, in the coming years and decades become increasingly obvious that robots walk amongst us. At present there are some 1.5million domestic robots. Known as Roomba and Scooba, these robots are about the size of a small dinner plate and will happily wheel around a room either vacuuming your carpet or mopping your floors. Okay, not quite the submissive android of TV, but the technology used by these robots is extraordinary; Set them down on any piece of ground and they will survey the area by trial and error, creating a virtual map of the floor plan; next time the robot is set down it will adhere to this floor plan, ‘knowing’ where obstacles are and how to avoid them. These technologies are not just for the domestic situation, but are of benefit in bomb disposal and isolating the survivors of earthquakes.
There seems to be a notion that an effective robot must be human-like, should stand like us and act like us, in essence it should be us, but an us we can command and dominate. The problem with this theory is that we have evolved over hundreds of millions of years and that for each evolutionary ‘success’ many species have fallen along the wayside, but still we try to impersonate evolution by building humanly biased robot and, it must be said, to some success. Honda’s Asimo robot is wonderfully humanoid- it has flexible joints, doesn’t seem to struggle when walking, but has a gait that is regimented and plodding. Fortunately there is an answer to this, human legs work in a pendulum type way, there is very little muscular involvement once we are on the move, instead we allow gravity to do what it does best.
Based on this idea, Tad McGeer of Simon Frasier University attempted to produce the whole walking step using what is an almost passive process. The Cornell walker uses passive walking strides more like a human than Asimo and uses up to 15 times less energy than its motorised counterpart. There are flaws in the human based idea, as passive walking only works on a consistent terrain, thus the robot couldn’t really change from gravel to carpet without toppling over. Passive walkers cannot control their direction either, so once on the move, the robot can’t turn around and head somewhere else.
Overcoming cumbersome walking is one thing, but if a robot is really meant to be humanoid then it should be able to talk like one. Waseda Talker in Japan is one of the first robots to use a mechanically synthesised voice. Compressed air is forced over a synthetic vocal tract and tongue which, coupled with a contortion on the mouth, allows the robot to form sounds without resembling the BT talking clock. This technology is still in its infancy but if successful the long term implications are immense. Initially we would have robots that, when the robot brain catches up, have the potential to talk to us, but we would also have the means to allow people with damaged vocal chords to talk again. In the past six years scientists have managed to form all 50 Japanese speech sounds.
So robots can walk and talk, but do we really just want a mechanised Mrs. Haversham wandering darkly around college halls? Robots without hands lack the essential dexterity to perform all of the tasks required of them, so robotic limbs are a key area of research. Despite being massively complex, scientists at MIT have planned to have a robotic limb with the dexterity of a six year old child before 2020. The limb should be able to tie a shoe lace and turn a page without being programmed to do so. The robot will rely on over 150 sensors per finger and calculate the relative pressures and movements it is applying and combine these in order to achieve its goal.
The combined robotic idea will achieve a mechanistic humanoid, who can at least in theory walk, talk and act like us (providing that ‘electronic brains’ catch up with the rest of the machine). Despite all this, how do you actully make a robot live? You give it an organic component…such as a slime mould!
A bright yellow slime mould has been put in the pilot seat of a six-legged robot. The slime (Physarum polycephalum), which naturally shies away from light, controls the robot’s movement so that it too keeps out of light.
Klaus-Peter Zauner at the University of Southampton says the idea is to find simpler ways to control a robot’s behaviour. The mould uses a network of tubules to sense its environment and decide how to respond to it.
Researchers grew slime in a star shape on top of a circuit and connected it, via a computer, to the six-legged machine. Any light shone on sensors mounted on the robot were used to control light shone onto one of the six points of the circuit-mounted mould, each point corresponding to the limb of the robot. As the slime tried to get away from the light its movement was sensed by the circuit and used to control one of the robot’s six legs. The robot then scrabbled away from lights as a mechanical response of the mould.
Biology is already influencing the evolution of robots in other ways. Chris Melhuish at the Bristol Robotics Institute has developed robots that generate power by consuming flies, so at least our mechanical friends will prove worthwhile for one thing