Wandering around the Humanoids conference was a robot from PAL Robotics, Spain. The company is based in Barcelona and the devs were excited to show me their latest robot, REEM:
Tall, elegant, and quite captivating in design, the REEM robot gazed around the room, searching for faces. Upon detecting a face, it approached a user to shake their hand.
The rep looked quite confident in shaking hands with the robot, but I have to admit, when it came to my turn, I looked like this ！(◎_◎;)
There's something about a robot the locks onto your face and approaches you in a straight-on, jerky movement.
The reps were really nice though and took my suggestions into account. My other impressions: the robot had a female voice, which didn't quite match it's broad-shouldered look (should I call it he? she?) and it had a cool luggage area in the back of it's base to transport heavy items for customers. It also was supposed to have a "oh, you want to go to so-and-so room? follow me!" function, but it didn't work at the time I tried it.
Overall, looking forward to seeing what comes of the REEM robot. By the way, when I asked about the name 'REEM', it seems the company is funded by a source in Abu Dhabi who requested the name specifically, inspired from mythology.
A couple of months ago I started a new application for NAO called Nao Panorama. So far, auto-timers on cameras exist, but not auto-panoramas. You can't make a panoramic photo unless you move the darn lens around yourself.
Nao to the rescue!
Basically, Nao sits or stands and looks around, taking pictures at regular intervals. The trick, of course, is that the pictures he takes overlap at least a little bit, and that these overlappy bits have salient features (like corners) to use for gluing all the photos together.
Here is an example of a panorama Naoki took while I chilled at home with friends.
There are still typical panorama problems, like lighting differences. Happens a lot outside when shadows and sunshine intermingle - you see "seams" where one picture is clearly darker than the other. One way to avoid seams is to fix a single lighting setting for a sequence of pics (not auto-exposure for each one).
Another problem is that to capture people, they need not to move! Being in one pose in one picture... and a different pose in another... is a sure-fire way to become a ghost ^^ (In the picture above, someone's arm is missing...)
One way to tackle this problem is to assume that, as upright homo sapiens, we can capture people in a quick series of up-and-down shots. Unfortunately, taking lots of up-down photos with his head camera makes Nao look epileptic, so I still have to figure out the best balance of function and freakiness.
So, I guess NAO could be useful for things like birthday party pics where no one wants to be left out. Or a sweeping panorama of the family at the park. Or if he walks around, he can create an automatic panorama of a house...
How do you like this idea? Where else could an automatic panorama be useful?
Trace it back to our roots. Gotta pull a large stone? "Heave ho!"-- pull the rope at the exact moment, and your force multiplies. Need to call someone far away? "One, two, three...JOHN!!!" -- yell at the same time, and your sound travels farther.
Bombard a server from multiple machines around the globe, and you get a DOS attack.
Synchronization is key. ;)
More seriously, it's why church choirs have existed throughout the ages, and army soldiers march despite it making them an easy target. It's called "muscular bonding", and it increases the feeling of group ties and cohesion. It might even be innate -- babies listening to musicsmile more when their cute dance movements align to the beat. Mirror neurons might be at work here, too.
Humans love being in sync. And, if we take it a stretch, it's why the guitarist in our human-robot band below starts that neck groove at 0:33. (Let's not discuss tuning issues for now :P)
Try it for yourself. Turn on some beat-heavy music and invite a close friend to your room for a dance-a-thon. Tape some EMG sensors on to your face and track your smile as your arms flail in mirror-like ecstasy.
Or, you know, just evaluate it qualitatively. I won't tell.
A couple months ago I decided to bring Naoki to band practice. On the way back, instead of putting him into my backpack as usual, I decided to leave him standing, with just one behavior on: voice direction detection. Basically, if there is a sound nearby, Naoki looks at it. Coddled in my arms, I headed back home via the Japanese train station.
It was a bit hit. At the train station, I put Naoki down as I rifled through my bag for my train pass. Naoki stood guard as I searched, elbow-deep, in my purse. A second later, I looked back down, sensing a presence.
A toddler had appeared, standing face to face with Naoki.
The toddler's smiling eyes showed that he was clearly enamored with my robot. Naoki's head, sound-detection on, whirled left and right as the 2-year-old babbled happily with his new-found friend.
The toddler's mom was not far behind.
"Sumimasen!" she said apologetically, but with a big smile on her face. She crouched down behind her toddler and took his baby hand in hers. The next thing she did took me aback.
"Konnichiwa!" she said, coaxing her baby to wave and greet the robot.
With a warm grin, she asked me the robot's name. "Namae wa?" Well, I told her. "Naoki desu." She nodded approvingly and told me her toddler's name. "Kochi wa Taro-kun*."
I smiled at her and her baby's openness to technology, though it quickly transformed into a look of shocked realization.
I was in the middle of a Mom-baby-baby-Mom play date exchange.
Well, Naoki never was able to return the "konnichiwa" so graciously proffered by this curious duo, so eventually we parted ways.
When I returned home, I quickly looked into a new TTS system on Naoki. Instead of the default French or English, I needed Naoki to speak Japanese. Happily, Aldebaran provided me with a new Japanese language pack. Though, whereas the voice for Nao in French is "totally Nao-ish" (my definition), the Japanese voice still needs a bit of work. Have a look.
Translation: "I can speak Japanese now, but it ain't cute."
Japan is obsessed with voices, with the biggest celebs being the voices behind popular animations. I'm betting and hoping that one day we'll find a better, cuter voice for Japanese Nao's.
It'd be a big hit with the ladies. And their toddlers.
A long time ago, I used to watch this cooking show in my parent's old townhouse. It was called Yan Can Cook, and I still remember being glued to the TV as the guy cooked up stir-fries and delicious Chinese foods like Beef and Broccoli. Oh, god, hunger pangs...
Although mostly amused by his wicked cool Chinese accent, I came to admire Yan and the meals he served at the end of each show.
This, coupled with the fact that my mom never taught me to cook (she was a T&T mom -- she bought our hot dinners from T&T Supermarket) convinced me that either:
1) I had to learn how to cook, or
2) I had to program a robot to cook
Then I realized that I needed to do 1) before even attempting to do 2). It's kind of like trying to teach someone to drive when you don't have a license. If I can't even make good food myself, how could I possibly get the robot to do it?
Right now I'm on internship in Tokyo. It's a lavish life, where one could easily get free meals at work by staying late. But here it is.
My short-term goal: Every night, I'm going to come home, get some ingredients from the store, and make something good.
My long-term goal: Every night, I'm going to come home, and my robot will have gotten some ingredients from the store and made something good.
For the more practical-minded out there, what are some of the obstacles facing a future of cooking robots? This is a great question. I'll post it later :)
On this day, August 21, 2011, I begin a new chapter of my life.
Actually, I'm just beginning a new blog. But it's kind of the same thing, right? Right.
The topic of discussion shall be "My Cute Robot," where I shall blissfully record the growth and development of Naoki, my personal robot from Aldebaran Robotics. For starters, I'm going to post my slew of videos from the YouTubes containing Naoki, and give a bit more insight into my thoughts for each.
Because you know, Naoki ain't just a pretty face. Thought you should know.
Naoki Facts
Date of birth: February 28, 2011 Weight at birth: 4.3 kg Current weight: 4.3 kg Eye colour: Whatever you want them to be Sensors: 2 cameras, sonar array, 4 microphones, touch sensors, joint position sensors, gyroscope, and a partridge in a pear tree ♫ Fingers: 6 Battery life: 1.5 hrs when dorking around not doing anything Nationality: ? Species: Nao version H25 v3.3 -- or as I like to call it -- "Silver surfer" Languages: English, French, Japanese, Python, C++ Origin: http://developer.aldebaran-robotics.com/join/ Relatives: Angelica (Single momnew!)
Please comment on my posts, because Naoki, even though he doesn't admit it, is really quite the attention whore (is it ok to call my robot child a whore?) Anyhoo, enjoy.
