Kia ora everyone!
So, Rob and I have been back in Melbourne a month and have been a lil busy. But its been fun. 😀
So briefly, we’ve been to Brighton Beach (kinda like Mission Bay of Melbourne), but it was so windy there were more kite surfers than sun bathers.
We also went to the Australian Open, hoping to see some tennis, but it poured with rain and we didnt see anything.
But the successful thing we did do is go to Sydney for a weekend. Had a *blast.* Admittedly, it was for a convention for Anothony Stewart Head who played Giles in Buffy. 😀 He was over in Oz with the Little Britain tour. Really nice guy. Here’s a few photos.
Tony giving the guests a small concert.
The group of us who went for a breakfast with Tony the next day. Big ass cruise ship in the background.
But we also did the touristy thing in Sydney….
Rob and our mate Carissa on a ferry boat over to Manly – a very surfy but busy beach.
So, all-in-all has been fun.
But now, for the masters
So, not a lot of progress, but my project has scope now. Before Christmas I felt like I had to do 3 PhDs worth of work, but thankfully now its confirmed in its scaled down form. I have a full Masters load instead. This is system is only meant to only be a concept system, so definately less work.
My topic, basically, is Resolving Ambiguities in Human-Robot Interaction. So, if we’re recognising the speech someone says and we’re not sure if they said ‘blue mug’ or ‘blue rug’, we can use our object recognition to see if (or maybe we’ve already seen that) there is a blue mug, but no blue rug. So that resolves that ambiguity in the recognition.
So the system will be a balance between speech, gesture and object recognition. Right now, I’m working on the gesture recognition, but I’m having trouble getting the system to find the arm and hand well. Am currently using samples of my own skin colour to teach the system what colour to look for. Not perfect, but working. Needs refinement. The system will only have 5 different gestures in the end – just a taste.
Next I’ll get a speech recognition system going, where it’ll only recognize about 20 words (keys verbs and nouns). And from that Wakamaru will be able to answer questions or complete tasks.
Finally I’ll get the object recognition system that I was working on last year working. And it’ll only recognize, say, 10 objects.
I have remember to do lil write ups on each section as I go, so then I dont have to write as much when I get to the thesis.
So, my Masters is meant to finish in March/April next year, though I might be able to apply for a short extension. Will see how it goes.
Hope everyone is well at home. Rob and I will be back in April for our birthdays, so might see a bunch of you guys then.