The developer of Honda’s cute android, Asimo, insists that the robot is here to take the drudge out of our lives – not to take over. Jon Excell reports.
For the past year Asimo, Honda’s walking robot, has been on a whirlwind tour of Europe, wowing crowds at technology shows and corporate events with its stair-climbing capabilities and dance moves.
It is tempting to view the robot simply as a clever marketing tool, and as a sophisticated showcase for Honda’s technical skill its impact is undeniable. But the diminutive android is much more than an impressive branding exercise. Prof Edgar Korner, the company’s robotics and artificial intelligence (AI) supremo, insists that Asimo represents a key step towards the era of the domestic robot.
Korner, head of Honda’s European Research Institute, is not a man given to hyperbole. He is candid about the limitations of current technology and speaks in the measured tones of an engineering realist. Because of this, when he talks about domestic androids liberating humans from the day-to-day drudgery of domestic chores he is all the more persuasive.
However, while Asimo is extremely impressive – its movement is quieter, faster, more fluid and more eerily human than one would expect – there is clearly some way to go before this vision is achieved.
In the immediate future Korner’s team is working hard to speed up Asimo’s movement and make its co-ordination more fluent. Currently, for example, the robot pauses briefly when switching between walking and climbing stairs.
The next generation, said Korner, will be able to move faster than Asimo’s current one mph, will be able to locate and recognise people and will be equipped with vastly improved speech-recognition technology.
In the longer term, Korner claimed, it is the technologies that we broadly define as AI that require the most work. ‘Asimo is a marvellous walking machine, a masterpiece of engineering,’ he said. ‘But the next stage is to enable it to develop the ability to ‘think’ for itself, to an extent where it can get on with its chores without bothering its owner.’
Despite some huge advances, AI technology is still at a very early stage relative to where it needs to get to. ‘We’re at something of a dead-end where the technology only works in a very restricted environment under highly predefined conditions,’ said Korner. ‘That’s not what we need if we want to make the robot freely interactive in a human environment.’
The further development of AI will, claimed Korner, be made possible by ongoing advances in the understanding of human and animal brains. It is this complex relationship between human and machine intelligence that is Korner’s particular area of expertise. He explained that the key to developing complex robotic systems is to learn from our own evolution, and to mimic the evolutionary development of individual capabilities.
Korner is fond of the way evolution’s ‘optimisation’ process led to tremendously efficient and effective biological systems. Crucially, he said, this did not happen in isolation – vision, intelligence, mobility all evolved in parallel rather than one by one. The same should happen for robots, said Korner.
By following this process Korner claimed that we should have domestic androids capable of moving about and performing rudimentary tasks in a restricted environment within the next five years. A robot that can cope with a completely unpredictable environment, however, is at least 10 years away.
In the shorter term, technology developed for Asimo is already having some interesting spin-off applications. Some of the mechanical advances are aiding the development of artificial skeletons for the disabled, and Honda’s work on machine intelligence is now being used to develop an accident-prevention system for cars. Such a system, said Korner, would be able to make split-second decisions allowing it to remove control from the driver to avert a collision.
However, Korner predicted that the development of systems in such safety-critical areas will be fraught with legal issues over how much control is delegated to them. The same issues will apply to the emergence of robots. ‘If we want it to be really useful as a service robot or a companion then it must be granted some degree of autonomy, and this, I expect, will be a step-by-step process, because we first have to establish some kind of control. How much responsibility we delegate to robots is a key aspect of Honda’s research,’ he said.
Some have claimed that there is a sense in which humanoid robot development – and more specifically AI – occupies a similarly ambiguous moral space to genetic engineering or nanotechnology, with scientists developing technology that has the potential to completely change the way we think about the world.
Korner does not agree. ‘From the point of ethics Honda was very careful to stress from the beginning that this is a machine. This is not intended to copy a human. The message is that we don’t want to copy humans, we want to create a useful machine for serving humans.’
It is not an entirely convincing argument when you see the robot in action at London’s Science Museum, the latest stop on its UK tour. Asimo’s cute appearance and overtly playful gestures – such as chicken impressions and surfing actions – are clearly useful in displaying the potential of the technology. They also do nothing to stop people from indulging in a spot of robot anthropomorphism.
Korner said the reason for this approach has much to do with social acceptance. Asimo’s predecessor, the P3 – a 1.6m tall storm trooper lookalike – would be a pretty terrifying presence in the average family home, even if it was just doing the washing up. Asimo, at 1.2m, looks far less threatening.
Another key factor to social acceptance, and an important area of Honda’s research, is security. The domestic robot of the future will be entrusted with a great deal of information unique to the user. It will, for example, be able to connect itself to the internet to carry out online banking for its owner.
But even if the robot is safe, fun and useful, is there anything to suggest that people will want it? Korner thinks so, and claimed that the market for domestic androids will be ‘really huge.’ He pointed to the trend in Japan for robot pets, which are no longer marketed simply as expensive toys but are also sold to pensioners as both companions and protectors. Able to respond to basic stimuli, these machines will even alert the emergency services if their ageing owners go quiet on them.
And then there is the inevitable question for every engineer working in the robotics field: what about the ‘Terminator scenario’ – the Hollywood-fuelled fear that robots will eventually become so intelligent that they enslave the human race?
Korner’s response is as unequivocal as you would expect from someone devoted to this area of technology. ‘Any tool made by humans to make life easier can potentially be turned into a weapon. Should we stop making knives? There are easier ways to kill people than to develop autonomous robots, and I don’t think it’s a reasonable fear that humanoid robots will dominate and control the world.
‘I expect robot development to set humans free to indulge our creative arts and our capabilities to deal with creative processes. By overcoming the limitations set by the limited capacity of our brain for storing information, we will be able to develop new areas of our creative designs and develop systems that were previously impossible.’
And to show that the robot visionaries can give as good as they get, Korner had a broadside of his own. To worry about robots enslaving the human race, he suggested, is ridiculous ‘when we should be worrying about the fact that half of humanity is already enslaved by trash culture, trash media and trash advertising.’