Assorted Cool Technology - 2012-04-03

Posted on April 3, 2012

Time flies when you’re learning fun stuff.

Having had a blast recalling what little I’d learned about AI, years ago in university, with Thrun’s and Norvig’s “Introduction to Artificial Intelligence” and also exploring the–until then–unknown science of Machine Learning with Ng’s introductory course a few months ago, I’ve now been expanding my knowledge about robotics, specifically self-driving cars (but learning lots of generally applicable stuff on the way) with Udacity’s CS373–Programming a Robotic Car (again by Sebastian Thrun).

The CS373 course is just about to finish and I now know more than enough to tackle a lot of projects I’ve dreamed up in my mind and also to finally get some other projects I’ve already tried to implement to fully work. Things get a lot easier when you know the maths and the algorithms, as opposed to simply throwing guesses at the problems.
What I lack is the tools, workspace and inclination to delve deep into the hardware side of things, but luckily I have friends who are more of the hardware and tinkering persuasion and so the potential for collaboration is high.

On a slightly different note, I’m also going through Coursera’s Game Theory class (by Jackson and Shoham), just because it seems really interesting and maths is kind of fun.

But getting back to the cool tech mentioned in the title of the post, this week we have…

Things that see

MIT Media Labs have been busy developing a camera that can see around corners. There’s a video that explains the process in simple terms which is actually quite interesting to watch.
It uses lazers! So it’s cool, right?

Also in the “I can see you!” department, has some rather nifty software that is really good at not only detecting faces and the mood of the person, but now also at guestimating the person’s age.

And then it gets creepy fast, when Hitachi Kokusai Electric starts selling a system that is capable of analysing 36 million images per second to detect and match people’s faces, with quite a bit of flexibility regarding the person’s pose and the size of the picture.
The fact that the system seems to be extremely well thought out and allows you to immediately jump into the relevant point in the video stream being analyzed and also to search multiple possible matches for the same person throughout the timeline can be viewed as a bit disturbing.
Watch the demonstration video for a glimpse of what it can do.
By now we should be more than aware that privacy is dead but this still creeps me out a little.

The robots keep coming!

Ever worried about our health, the helpful robots are now ready to try to prove their mettle on the Hospital floor.
Of course this is early days and this first trial will mostly let the developers of the robot know how much and in what ways it gets in the way of the doctors, but it is a necessary step towards taking the drudgery out of the hands of junior doctors, allowing them to focus on more important aspects of their development as professionals.

And now for the really cool and shinny piece of tech of the week: the Mighty morphing hexapod bot is back!
This spheric hexapod has been making the rounds for some time now and it has been learning and getting new abilities at a nice pace, until it reached what you can watch in this video.
That doesn’t mean that the development is done though, after all, there’s so much potencial with this lovely toy!