An Oxford University team will be one of only two teams* from the UK to compete in RoboCup 2008, the ‘Robot Soccer World Cup’, in Suzhou, China 14-20 July.
Whilst the long-term aim of the tournament is to produce a side of autonomous humanoid soccer players that can thrash the most recent human World Cup-winning team, right now its leagues and challenges are a showcase for new ideas in robotic hardware and software.
'The most spectacular events are the robot football games with physical robots, with the leagues split according to the size of the robots, and whether they use wheels or legs,’ said Stephen Cameron, one of the three researchers from Oxford’s Computing Laboratory attending. ‘Many of the issues being tackled here are the serious engineering problems of building robots capable of moving a ball around, whilst being reliable enough to still be operating at the end of the game!’
Away from the physical kickabouts robotics is taken to the limit in the Simulation leagues. Stephen explains: ‘These use simulated robots within computer programs to push the boundaries of robot intelligence without the constraints of what is physically possible today; the computer programs written here will find their way into the physical robots of tomorrow.’
Stephen and his research student Jie Ma will be competing in the simulated wheeled and legged robot competitions whilst another of his students, Julian de Hoog, will be collaborating with a team from the University of Amsterdam in the RoboCup Rescue competition, in which a simulated robot tries to identify and prioritise casualties within a collapsed building.
‘The simulations we use closely resemble real situations and the virtual robots are based on real robots,’ Julian tells me. ‘The long-term goal is to develop robots to crawl into spaces within a building that has collapsed, spaces that, because of the size of the hole or the poisonous gases or heat involved, wouldn’t be accessible to humans or dogs.'
Up to fifteen different models of robot exist in the RoboCup Rescue simulations: it’s up to competitors to get them to work together to perform tasks guided by the sorts of limited sensory data and communications available in a real disaster situation.
‘I specialise in developing strategies to help robots explore these environments,’ Julian explains. ‘The eventual aim is to get robots to learn on their own how best to spread out and search an unknown environment for survivors in the most efficient way.’
It’s likely be many years before real robots routinely do lifesaving work alongside humans – for one thing, we need to learn to trust our metallic friends – but within a decade expect to see some amazing rescue machines.
[* in the senior competition, good luck UK juniors!]