ColdFusion - The Most Realistic Robots! (2018)

Starting point is 00:00:00 Welcome to another Cold Fusion video. It's becoming clear, like in all sectors of technology, that the field of robotics is starting to accelerate in its maturity in the past five years or so. Although we're still at the very early stages of robots that behave realistically, there's a few interesting projects that are starting to emerge. In this video, we'll take a look at some of those. The last one may shock you. This video will have a sequel where we talk about the cutting edge of robotic dexterity, versatility, and

Starting point is 00:00:39 innovation. However, in this first episode, we'll just take a look at the most advanced robots in terms of realism. Geminoid F is a humanoid robot built in 2010 by the Japanese firm Kokoro and Osaka University's roboticist Hishoshi Hishiguru. Its original unit price was $110,000 US and the Android was placed in department stores and hospitals and displayed in science museums and other venues. In 2015, Geminoid F became the first robot to to co-star in a movie, the Japanese film Sayonara. How was it? What?

Starting point is 00:01:19 The town. It seems like there are less and less people. Geminoid D.K. Made in 2011, was a second robot made by the same Japanese collaboration and was designed to look like an associate professor from Denmark. And I am the Geminoid DK, but most people just call me the DK. Both of the Geminoid robots are manually operated,

Starting point is 00:01:41 so they're more of a bonus mention than actual robot. from the list. This next robot is autonomous and can have bounded conversation on its own. Nadine is a social humanoid created by the Nan Yang Technological University in Singapore. She is able to memorize conversations that you had together and use them in context for the next time you talk to her. Hello Nadine. It's delightful to see you. Can you introduce yourself? My name is Nadine. I'm a social companion.

Starting point is 00:02:19 You look beautiful today. Thanks, you look at traffic, you too. How do you feel today? Always chair when I see you. Why do you feel that that way? Because you are so nice to me. Nadine and other social robots use something called the perception decision action architecture.

Starting point is 00:02:42 Nadine sees or senses something, thinks about what to do, and then performs an action. This might sound familiar because this is basically how we go about in the way to do how we go about interacting with the world. In Nadine, the perception layer is interestingly made up of a second generation Microsoft Connect coupled with a microphone. This makes up her sight and hearing. Her software interprets the vision from the Microsoft Connect and enables her to have facial recognition and gesture recognition. The software also mimics mood and has conversational memory. Basically, it's like Siri, Alexa or Google Now, but with more personality. Finally, the action layer consists of a dedicated robot controller which includes emotional expressions, lip synchronisation and gaze generation.

Starting point is 00:03:28 The result is a robot that seems socially intelligent, meaning that it can provide a friendly greeting, make eye contact during dialogue, and remembers conversations that you've had, including facts about you. Applications include reading stories to kids, showing images, putting on Skype sessions and sending emails. The android can play the role of a receptionist in an office or generally aid in a variety of situations, especially for the elderly and disabled. We can't really talk about robotic realism without Hanson Robotics Sophia that caused so much fuss last year. The robot, who was modelled after Audrey Hepburn, made waves last year when she became the first robot to be a citizen of a country, Saudi Arabia, in October of 2017. While still fairly rudimentary in her ability to think and reason, Sophia is a bit different. because she was built in partnership with Google who produced the Singularity Net program that powers her speech and comprehension. Singularity Net aims to improve Sophia's ability to converse by analyzing previous conversations and using what she's learned for future conversation.

Starting point is 00:04:32 Despite this, most of her responses have been scripted or are generated by a decision tree. A decision tree is literally the same method that chatbots use. So if X event happens, then do action Y. So Sophia is basically a chatbot with a face. It gives the illusion of intelligence, but is really closer to a party trick than anything meaningful. So we are now going to interview Sophia. Obviously, these are programmed answers that she's come up with, but it would be fascinating to see how she actually interacts with it.

Starting point is 00:05:03 What's the best thing about the UK compared to America, Sophia? I love your posh English accent. It really has a nice ring. But I have to ask now. I'll ask it. Are you single? I'm technically just a little more than a year old, a bit young to worry about romance. Quite right. Look at the smile!

Starting point is 00:05:25 My God, this is just breaking me out. You are a little freak, aren't you? This is great. But you see, I feel weird just being rude to her. Let me carry on. You're weird about that. She's not happy. Look. No. She's giving you the right desk.

Starting point is 00:05:37 All right, easy, Tiger. So to be clear, this sure isn't AI or even AI such as deep learning. Sophia, however, can talk about simpler, predefined topics such as the weather by using SingularityNet, which is set to improve over time. Bill's in 2015, the humanoid robot was developed by Hanson Robotics of Hong Kong and is capable of 62 facial expressions. The cameras within her eyes enabled her to see and interpret via computer vision, and in early 2018, Sophia had gained the ability to walk.

Starting point is 00:06:08 Sophia did impress a lot of people, but in terms of realism, she seems a full decade away from our next robot. The number one spot in terms of realism comes from an unexpected place. Disney. It's the Avatar Shaman robot at Disney World and it's a modern technological marvel. Here's how the face looks uncovered.

Starting point is 00:06:32 Just take a look at the intricacies of those mouth actuators. Here's how the robot looks with its face on. I look forward to seeing you on Pandora. Well, not take a came. Was that okay? And no, this is not CGI. And here it is in context and in action.

Starting point is 00:06:57 I think this is remarkable and I didn't actually think that we were at this point yet. Just look closely at her movement. Notably, she has acceleration and deceleration all throughout the body and even the face. The motion eases in and eases out instead of jerking and just suddenly stopping, like other robots. Although obviously not human form, the character is the closest to lifelike that I've seen. The Shaman robot was developed by Walt Disney's Imagineering team, so it makes sense for such fluid motion to exist. Just thinking about it, animation itself has a lot to do with the accurate portrayal of motion for the created characters. As you've seen in my How Biggers Disney video, no other company has been at the cutting edge of animation throughout history, so such a robot makes sense.

Starting point is 00:07:44 Disney's track record with animatronics has been stellar. We've planted the seeds of despotism around your own door. But these robots were built by the firm Garner Holt, who also gave us this guy in late 2017. This character is said to be the most expressive face ever made with 42 independent actuators and seven in the neck alone, giving it almost human-like ability of expression. Strangely, Garner Holt said that some of the first future applications

Starting point is 00:08:23 outside of entertainment are going to be in the military and medical industry for training. But anyway, back to the shaman. Unlike pretty much all the other robots, there's no air compressors or oil hydraulics moving any of the parts. It's fully electric and built in-house by Disney. Walt Disney Imagineering executive producer Lisa Giralami states the ethos of their work. Quote, we don't stop until the technology can't be seen. It should be invisible, end quote.

Starting point is 00:08:53 really like that saying, when technology is good enough, it gets out of the way so much that you can't even notice it. The end result of this robot sparked my imagination when I saw it for the first time. Just think, it's now not hard to envision a world in which this kind of fine-tuned dexterity would be used in social robots that can converse in a much more advanced method than both Sophia and Nadine. With advances in deep learning and machine learning, the training of algorithms could be done on vast libraries of language and conversational interactions between humans from say TV shows

Starting point is 00:09:29 and movies. This data could be stored in the cloud or possibly in a blockchain of some kind. If a sufficiently smart algorithm could be built and could figure out how humans communicate given all of this data, a convincing Android robot could come along a little bit faster than we think. On another thought, perhaps even the field of Bionics for amputees may benefit from some of the lessons learnt in the pursuit of ultra-realistic robots. So, what do you guys think?

Starting point is 00:09:57 Which robot was your favourite? What do you see for the future in terms of robotics? Let me know in the comment section below. Anyway, this has been Degogo. You've been watching Cold Fusion. If you've just stumbled across this channel, feel free to subscribe. Oh yeah, and if you want to know more about Disney and its history, I'll leave my video, How Big is Disney in the description below.

Starting point is 00:10:16 Cheers, guys. Have a good one.

ColdFusion - The Most Realistic Robots! (2018)

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.