It may seem surprising, but it is the first category which is the breakaway, avant-garde one, and is garnering sponsorship from forward-looking corporations such as British Telecom and Matra Marconi Engineering.
It is typified by the robot "insect", which has one simple objective: to get as much sun on its "back" - that is, its solar panel - as it can. The descendants of such robots could one day be prospecting on Mars or other planets, autonomously operating in environments too hostile for humans.
It was one of a gallery of robots being shown off yesterday in an exhibition at the Brighton Metropole Hotel, where the fourth European Conference on Artificial Life (ECAL) is continuing this week.
Coincidentally, yesterday the Honda corporation of Japan released a picture of its P-2 robot - a 6ft (1.8-metre), 210kg creation which is claimed to be a world first. It can climb stairs, balance itself if pushed, and keep itself upright on uneven terrain. But that is the sort of crude mimicry that many gathered in Brighton would find uninspired - a solution to the problem of moving about that harks back to 1950s science fiction.
Instead, robotics is now embracing evolution and biology with open arms. "One idea is like Darwinian evolution for the control systems of robots," said Phil Husbands, a co-organiser of ECAL who is also a Reader in Artificial Intelligence at the University of Sussex. "You create an artificial `selection pressure', and it turns out that it requires very, very minimal computing compared to the traditional robotic approach."
The "traditional" view does have its place, he noted. "If you know exactly what you want - say for car-spraying - then you would be crazy to try to evolve a system. Traditional control theory can cope with that fine."
The "evolutionary" approach to robotics lets the software of the robots' control systems "breed" according to how well it does in a task, so that the best-adapted will be chosen, and potentially improve on human-designed solutions.
"Companies like BT believe that learning more about biological systems will help with engineering systems," said Dr Husbands. "After all, biological systems are far more complex than any that we have made, but they work very, very well."
He believes that a "bio-robot" designed according to evolutionary principles could be sent to unpredictable environments and survive.Reuse content