Shutterstock | maxuser
We began our day at ESOF 16, a conference celebrating the crowning of Manchester as the European City of Science, listening to an interdisciplinary panel discuss developing advanced robotics for interactions with humans.
The development of robotics has advanced significantly since they were first imagined, in their current form, in the 1920’s sci-fi classic Rossum’s Universal Robots which introduced the masses to humanoid automatons, programmed to help and assist as butlers. However, the story takes a dark turn when the robots turn against their human masters, ending in chaos and destruction.
This story taps into an anxiety that still remains today. We want robots to act human as it will allow us to communicate more effectively with them, and for them to fit more easily into our everyday lifestyles. However, the more humanoid a robot is, the more uncomfortable they appear to make people feel.
Karel Čapek's genre defining play, Rossum’s Universal Robots, explored the uncomfortability present in human robot interactions
There is something unsettling about machines that appear as humans. This effect has been documented extensively with the most famous example being the uncanny valley, if a robot’s design is visually similar to the human form people are more likely to be scared by it.
The 'uncanny valley' describes how as a non-human object approaches a high level of human likeness people are more likely to be scared or repulsed by it.
This effect is unfortunate as robotics has the ability to revolutionise our society. Robots could be used for anything from the mediocre (no more ironing!) to the magnificent; such as surgery, education and aviation. Developing new ways of overcoming this comfortability barrier so we can use robots to their full potential remains a big obstacle.
The shape of robots is not the only area of concern too, the mechanisms by which robots interact with humans must also be fine-tuned for these interactions to be beneficial.
This panel discussed the ways in which robot cognitive programming must be based on human cognition to optimise the benefit of human robot interactions (more on this later), with a particular focus on using robots to facilitate the rehabilitation of disabled children.
The interdisciplinary panel included Friederike Eyssel, a professor of psychology from Bielefeld University, Dr. Marina Mara, a robopsychologist from Ars Electronica Futurelab Austria, Yukie Nagai, a professor of roboengineering from Osaka University and Giulio Sandini, a professor of Bioengineering from the Italian Institute of Technology.
Professor Sandini began the discussion by introducing us to the history of robots, demonstrating the leaps and bounds that have been made over the last century. This was fascinating in itself, humans have been having dreams (and nightmates) of robots since the time of the Greeks, but only now are we close to realising this dream.
Robots Helping the Blind
This introduction was closely followed by an exploration of the psychology of robot-human interactions, led by Dr. Marina Mara, who has been developing robots to help assist blind children to learn about the world around them.
Almost all rehabilitation treatments for blind people are geared towards adults, with little useful technology available for blind children. This came as a large surprise, especially considering that cognitive development can be impaired by the lack of this sense.
Part of Dr. Mara’s research has involved studying human-robot interactions with the aim of improving their quality. Dr. Mara has made some very astute observations that show that humans are more likely to trust, and show affection towards, robots which are imperfect. In her research she noticed that trust in a robot was significantly (and counterintuitively) increased if it occasionally made minor mistakes.
She hypothesised that the perfect nature of robots was something that made humans feel uncomfortable.
Disney hero Wall.E is an example of an imperfect robot, his clumsy nature meant audiences felt endearment towards him. Shutterstock | Nicescene
Developing Human Cognition in Robotic Systems
This led on to Professor Nagai’s research on developing robot cognitive behaviour so that their interactions with humans do not cause discomfort. Professor Nagai drew attention to the phenomenon of mutual understanding and anticipation which are a key part of human psychology and human-human interactions.
She demonstrated that humans have an innate capacity to help one-another, regardless if they are asked to help or not. This point was pressed home by videos of toddlers helping adults, who are struggling, by opening doors without being asked for help.
Experiments with altruism in children and chimps
An experiment that demonstrates the helping behaviour that is deeply embedded in human psychology. YouTube | johnnyk427
This behaviour is a direct result of the way our cognition is 'programmed' into our brain. Our brains have developed in such a way that we can anticipate another human’s intentions and have mutual understanding of their emotions. This is a cornerstone of human interactions.
The panel agreed that developing this cognitive model in robots is a key part of creating robots that are helpful and don’t make people feel uncomfortable. However, in order to achieve this, we need to develop our understanding of the psychology of human interactions.
Currently, there is a gap in knowledge in the field of human psychology between macro behaviour, such as helping others, and the micro activity that occurs as neurons signal to one another. Current research is working on building a model of human cognition which can then be used to provide a model for cognition in robots.
To create robots that display human traits such as helping behaviour, empathy (and the accidental error) scientists must develop electronic circuits that behave in a similar way to neurons in the human brain. Shutterstock | chuyuss
How Close are we to Advanced Robotics?
Whilst we are undoubtedly far closer to having the robot helpers dreamed of in our films and literature than we were a century ago the panel agreed that we are still decades away.
Robots have the potential to impact our lives in ways we can’t even imagine. But, it appears that to develop the robots we require, and robots people will want to interact with, we need to look inside ourselves a little more and understand what it means to be human.
This piece is part of our coverage of ESOF 2016, for more information on the conference and the talks we attended please see our overview article.
We're also in the process of conducting interviews with leading scientists from the conference. Watch this space!