Skip to main content

Disney developing robots to chat with emotion, sculpt and fabricate

Disney Research robots sculpt clay

Disney Research has well developed robotics and AI programmes. Recent publications show that they are working on new functions for robots: chatting with emotion, sculpting clay and fabricating metal forms.

We have already seen Spider-Man stuntronic robots that can perform acrobatic feats, and robots that can mimic human facial movements. Disney Research’s latest advances are in robotic sculpture, fabrication and realistic emotional chat. Read on to find out more, including how Disney robots become downright unpleasant in conversation!

Disney robots: sculpting clay

Disney researchers have developed robots that can sculpt clay.

The interactive design system starts with an input mesh and works out the optimum cutting path to for complex curved shapes like faces and eyes. Users can choosing from different sculpting styles. The finished piece is carved using a 6-axis robot arm.

Disney robots: fabrication

This piece of research looks at how robots can be used to fabricate complex, but structurally sound, metal frames. The techniques start with a model and break it down into sections that can be fabricated. The robot then makes the final piece by bending bars, cutting and welding.

Disney develops AI to chat emotionally

Disney Research robot AI chat

Perhaps the most intriguing piece of research is the developments in programming robot AI’s to chat realistically with humans.

In something called “Dynamic Emotional Language Adaptation in Multiparty Interactions with Agents”, researchers are attempting to create more realistic, empathetic and believable dialogue that is “not only relevant, but also emotionally appropriate and consistent”. The system tries to work out the emotional state of the human it is ‘talking’ to and then adapts it’s responses accordingly.

The paper says that “implementing emotionally-appropriate language has been shown to increase rapport with users, which is likely to improve both short- and long-term interactions, effectively building a relationship between the agent and the user” which ultimately “creates a conversational foundation of empathy and trust”.

Here are some sample conversations from experiments where real humans interacted with the system. In the first conversation the AI, Nobu, talks about a social conflict with its friend and asks for advice:

U1 Hello
AG Oh, I’m glad you’re here. My name is Nobu. What’s your name? [relief ]
U1 dt [ASR error]
AG Wow, I didn’t hear what you just said but I’m just glad you’re here anyway! [joy]
AG Oh look! Another human! What’s your name, Other Human? [joy]
U2 My name is [U2_name]
AG Can you believe how popular I am, [U2_name]? Two human friends in one day! [pride]
AG You know, I’m considerate, even in the face of tragedy, so I’ll ask you first. How are you? [pride]
U2 Oh I am doing alright
AG My presence alone just makes such a positive impact. [pride]
AG I really do care about my friends. How are you, Other Human? [pride]
U1 I’m doing well
AG That almost makes my worst day ever a little better! [joy]
AG It’s perfect that you’re here, because I have a huge problem. Can you help? [joy]
U2 No
AG You’re such a good friend I know you’ll help anyway. I did something bad and now I need to
lie to my friend. [love]

In the second, the AI simulates having followed the advice and it doesn’t seem to have gone so well. I’m not sure Nobu will be working at the Magic Kingdom with this attitutude!

U1 Hello
NA Oh great, it’s you. I’m Nobu. Who are you? [hate]
U1 My name is [U1_name]
NA Well, thank goodness you’re here. [relief ]
NA Seriously? You brought this guy? Ugh. What’s your name, Other Human? [hate]
U2 My name is [U2_name]
NA Bleh. Human names are just so HUMAN, it makes me sick. [hate]
NA You people show up on my bad day and now I’m supposed to ask you how your day is going?
Just tell me. [hate]
U2 My day is now worse for having met you
NA I don’t even really care what you said because everything is awful and I’m mad. [hate]
NA Of course there are two of you, so I have to be polite twice. How are you, Other Human? [hate]
U1 I’m doing ok
NA You are so insensitive to my feelings. You know today is my bad day. [anger]
NA Since you’ve decided to hang around, why don’t you make yourself useful and help me with a
problem I’m having. [anger]
U2 No we don’t want to
NA Thanks for nothing! I’m telling you anyway! My so called friend put me in a bad position and
now I need to lie to him. [anger]

Share this
Rachel Reed

Rachel Read

Rachel is Finance Director. She has a degree in engineering from Cambridge University and qualified as a Chartered Accountant at Deloittes in London. She worked in finance in industry for twenty years. She oversees our news and also manages our events.

More from this author

More from this author

Related content

Your web browser is out of date. Update your browser for more security, speed and the best experience on this site.

Find out how to update