Science

The shaky human robot team

In a seminar for the occasion of Delft’s 174th Dies Natalis notable researchers from all over the world presented their views on the future of robotics at TUD. The main message: it is not about dividing tasks between humans and robots, but about continuous cooperation to get the best results.

In the sixty years or so since robots have held the promise of combining physical and mental tasks at a (near) human level anxieties have remained constant, namely that at some points robots will take over control of human lives and societies. What has changed in the meantime is technology itself. So, while the anxieties voiced at the seminar ‘Intelligent robots: tools or team mates?’ sounded familiar, the current state of technology came with more soothing insights.

The voice of alarm was raised most distinctly by Dirk Helbing, author of ‘The automation of society is next: how to survive the digital revolution’. ‘Robots know how to learn and to design, so they will be able to design better robots’, Helbing said. ‘They will become our team mates and possibly our bosses. The question is what kind of culture we engrain in our systems. The Chinese already have a system that follows every step you take on the internet and gives you points for good behaviour. If you look at the list of the freest places in the world, none of them is an ICT development centre. I really think we need more democratic systems.’

Helbing’s warning was echoed in the lecture by Sebastian Thrun, the leading man behind Google’s driverless car: ‘Every human has to learn how to drive individually. When a driverless car learns from a mistake, all cars learn. The spread of robot learning is outpacing human learning.’

Yet Thrun posed as an optimist, noting how much humans will benefit from computers taking over their cars. They drive more safely and are likely to solve congestion and parking problems as the era of individual car ownership comes to an end.

MIT’s Tom Sheridan, old enough to remember the pioneering days in the fifties, reminded the audience that there are many shades between full robot or human control. Actually, there is no such thing as a fully autonomous robot, Jeff Bradshaw of the Florida Institute for Human & Machine cognition, claimed: ‘Just as there is no fully autonomous soldier or sailor, autonomy is something that emerges from the system and it is never static over time.’

Delft’s own Catholijn Jonker proved the point by showing a video of a robot teaching a child about diabetes. Control of the situation passed back and forth between the two in a natural way. Jonker compared the cooperation between man and computer with the symbiotic riding of man and horse, each doing what it is best to achieve the optimal pace.

Bradshaw showed his way of designing human robot communication by mapping in detail which information robot and human need from one another to function at their best. ‘Machines used to be dependent on humans,’ he said. “Then they were designed to operate independent of human input. Now we have to design for interdependency.’

In the end, the question of autonomy may have more to do with how we see ourselves than how we see robots. If we no longer drive cars have we lost something that defines us as self-reliant humans? The answer will differ from person to person, though all robots will probably render the same output.

Editor Redactie

Do you have a question or comment about this article?

delta@tudelft.nl

Comments are closed.