TED talks are short presentations on ‘Ideas Worth Spreading’ given to a live audience by a vast range of experts on a vast range of subjects. From future technology to identifying fossils, from statistics to stand-up (some of the talks are very funny), they can be found online on YouTube or on the TED website and I highly recommend taking a look for yourself. Here, I will be sharing my favourite TED talk that I’ve seen each week.
Guy Hoffman: Robots with “soul”
Guy Hoffman: Robots with “soul” (17:39) TED x Jaffa 17 Jan 2014
Guy Hoffman was inspired to create emotional robots when he watched Pixar’s ‘Luxo Jr.’ short film. The film shows two of the iconic Pixar lamps playing with a ball. He was amazed by the fact that the animators managed to put so much feeling into two pieces of furniture. Guy Hoffman was inspired by this and decided to move to New York to study animation. he spent half of his time at school, studying, and the other half drawing, frame by frame, a pencil animation.
While studying, he learnt two important lessons. Firstly, that if you want to give something emotion, it doesn’t matter as much about the way it look, but rather how it moves and interacts with other objects and characters. This is how the animators at Pixar managed to give two desk lamps emotion. The second lesson, told to him by one of his teachers, an animator who worked on the film Ice Age, was that an animator is an actor, not a director. His teacher suggested that he stood in front of a mirror or a camera and acted out the movements that he wanted his character to do, and then put them into the animation.
A year later, he was working at MIT, in the Robotic Life group, one of the first groups to study the relationship between robots and humans. While working there, he discovered that robots didn’t move in the same fluid and free way that he was used to in his animations. They were all too robotic. So, he then used all of the skills he had learnt while studying animation and used them to make his own version of the Luxo Jr. desk lamp. Hoffman then showed a video of his robot in action. It was a fully functional desk lamp that could move around independently. It followed him around the room when he went from his desk to a whiteboard and back again. Also, when he couldn’t find something, the lamp looked at it and focussed its light beam onto it. By doing this, he wasn’t creating a functional machine that simply shines a light bulb, it had become almost a ‘quiet apprentice’ helping him out when he needed it, but not interfering.
He then demonstrated how the lamp, with exactly the same physical structure could either seem caring and friendly or threatening and angry. He demonstrated this by showing a clip of the lamp interacting with two actors, one tired and sad, the other angry and violent.
At this time, Hoffman was working on his PhD and was studying human robot teamwork. As part of this, he was studying the engineering, the psychology and the philosophy of teamwork. While doing this, he noticed a difference between the way humans interact with other humans and the way robots and humans interact. He noticed that when humans work together, there is little need for communication or instruction. Everyone knows what they need to do and they do it as quickly as possible. WHen humans are interacting with robots, on the other hand, Hoffman described it as more of a ‘chess game’ the robot waits to see what a human will do, it analyses the human’s action, the robot decides what to do next, plans it and then does it. Then the human waits to see what the robot does and then responds similarly.
Hoffman acknowledged that this was an efficient way for a robot to work as chess is a game of logic, maths and decision making, just like a robot, which acts based on logic, maths and decision making. But Hoffman didn’t want to make a chess player, he wanted to make a doer, that quickly responds to external stimuli without considering multiple outcomes and consequences.
Hoffman then took a break from his PhD to study acting for a semester. While studying acting, he discovered the technique of method acting, where an actor improvises using their whole body, based on what their fellow actors are doing. He tried to implement this into his robots and tried to create a robot that will react with it’s whole body to what someone else is doing.
He then performed an experiment where he got volunteers off the street and got them to interact with his lamp robot, installed with two different brains. One was the calculating, ‘chess player’ brain but the other was the risk taking, ‘actor’ brain which didn’t act with all of the information it needed and improvised. Half of the volunteers were interacting with one brain, the other half with the other brain. Both sets of volunteers were performing the same, mundane task in which the human and the robot had to work together. The volunteers who were working with the ‘actor’ brain much preferred it, even calling ‘he’ or ‘she’. One of the volunteers said that ‘by the end we were good friends and high-fived mentally’. The volunteers interacting with the ‘chess player’ brain called it ‘it’ and one of them said that ‘it just felt like a lazy apprentice’.
Hoffman then decided to create a musician robot. He wasn’t the first to make a music playing robot, but all of the others were ‘chess players’, Hoffman created an ‘actor’ robot musician. he started with just four arms holding sticks, playing a marimba, which improvised in accompaniment to a pianist. Hoffman then added a socially expressive head to the robot which could bob in time to the music and could look at the musician and indicate when it wanted a particular chord to be played. He then took it a step further and created a speaker dock which tapped its foot and danced in time to the music.
Guy Hoffman is an incredible inventor and thinker. He creates robots that push the limits of what is possible in robotics and can be mistaken for having human emotions and can engage and interact fluidly with humans. Hoffman believes that ‘actor’ robots are less than perfect, yet are perfect for us. His talk was enlightening and showed me the possibility of a future with expressive robots.