Swimming

I decided to take up swimming in an attempt to make my cuddly body a little more rugged and manly. On Tuesday and Thursday and swam 1.25km each day. This means that this week I have completed a little over 7% of a cross-Channel swim, with a few days rest between 3.5%s. I don't know what all the fuss about David Walliams is about. At that rate, with strategically placed islands throughout the Channel, that served burgers and fried food, I could complete the channel swim in a breathtaking three and a half months, (ignoring currents and the fact that I suspect that the pool I go to is a tad warmer than the Channel.)

Invasion

I got home today to find that my room had been invaded by the evil ant horde.

They have been threatening an invasion for a few weeks now from the opposite end of the corridor, but they always seemed to get as far as the kitchen and stop. This was great for me, as they stayed away from my stuff in the kitchen, and I'm at the opposite end of the corridor.

Many people at the kitchen border had employed chemical deterrents, (ant powder at their doors) but I sat smugly at my end knowing I would have plenty of advanced warning were they to attempt an incursion to my area.

Sadly, they made it through the fire escape. Our weakest border was taken advantage of, and my room was the nearest target. They had conga-lined all the way across my desk, and even had the audacity to march across my keyboard when I found them.

Retaliation was swift and harsh, I immediately ordered the deployment of a HMS, (Hoover of mass suction) and condemned them all to the Gu-ant-anamo Bay that is a small Hoover bag, without trial. The bag is now being shipped to Cuba, where the world's press will frown on it for a week and then get bored.

I phoned for backup from the UN, (university nitwits) in the porters' cabin, who told me that it was an estates issue, and only the head of housekeeping, who wouldn't be in until tomorrow, was allowed to call them. God bless university bureaucracy!

Fortunately a friend of mine had pre-empted an attack and had an ant-bait trap. I traced back the pheromone trail, scrubbed a hole in it with soap and water for a while and then placed the bait trap on the side nearest their entrance. As we speak, it is swarming with the little buggers, all carrying slow acting contact poison back to queeny and the children, thus wreaking my re-wengy.

I still see the odd post-Hoover-apocolypse ant, wandering about on my desk, looking lost and bewildered. I'd like to think that he eventually comes across the red biro on my desk, realises where he is, falls to his knees, (all 6) and screams, "You Maniacs! You sucked it up! Ah, damn you! God damn you all to hell!"

Air con

Is buying an air cooler for £45 because my room is basically a sauna, and on the ground floor so I can't leave the window open, extravagant?

Discuss.

PhD Things

I haven't posted on my PhD status for a while, so I thought that I should give you all an update.

I've been reading A LOT recently and I think that I'm homing in on something which is interesting enough for me to spend the next 2 and a half years studying! I really wanted to do something with co-operative robotics, as I really want to better understand the processes by which multiple agents can give rise to more complex behaviours through interaction. That's fairly standard stuff, I've just bought a book on Evolutionary Robotics, as I need to start looking at existing algorithms for implementing what to me seems like a task which is annoyingly, 1) Too slow to implement realistically on an embodied agent, and 2) Too dependant on physical conditions to implement on an agent that is required to operate usefully in the real world. I'm guessing some middle ground, where the majority of the grunt work is done on a reasonable simulator and then some kind of tuning process to make it work on the embodied agent, though the pitfalls there are already obvious, (there may be much better solutions to the problem that can take advatage of real-world properties not fully encapsulated by the simulation etc.)

I'm just about to start reading a paper about the application of reinforcment-learning-based-machine-learning-algorithms to large groups of robots, for cooperative tasks. I'm hoping that this, combined with the book, and as many references as I can get from reading the book, and looking up the appropriate terms in journals, will provide me with the basis to start applying my own ideas, (or at the very least, know the "in" terminology for my ideas, so I can check if some other bugger has got there before me!)

Other than that, I'm still waiting for information that I asked a colleague for about a month ago, and he's never in when I go to his office, (dead??) and we're in conference season, so tracking anyone down whilst they're actually in the country is akin to hunting a specific ant in an ant colony that's run by a queen with a genetic propensity to producing offspring that look paticularly similar... in the dark.

I've written a few paper summaries here and there and haven't really started to structure anything into a formal lit review, but I'm hoping that with a bit of rephrasing of my summaries that I will be able to create it fairly easily. The only issue is that I may need to reread any paper that I choose to use, which I read at the beginning of this process, as I don't think that I was originally wary enough about avoiding words and phrases creeping in that could lead to accidental plagurism.

Other than that, I've been a bit slow on the coding front. I am still yet to play with the MS simulator, which I really want to do, especially now I've been informed that it DOES indeed work with the cheap-ass version of MS Visual Studio that the uni provides, and isn't in the "supported IDE's" list for the new system, (so I'm guessing that it means that they're not testing it on that IDE, rather than explicitly excluding it.) I'm also yet to play with the camera feed stuff that I wanted to. I really think that a nice image processing demo for the lab would be cool, and the experience could well let me create a nic framework for plugging such information into a controller.

Any hoo - that's it.

boB out.

Yet another test

Once again, I am attempting to make use of the blogging via email feature
of my blog.

The results tend to either be that the post doesn't go up at all, or takes
a few hours to go through!

Hopefully this will actually work!

Arguing

I had a row with one of my close friends yesterday, over something stupid. I was a jerk. My bad. Thought that announcing it to all and sundry might be cathartic. Anyone else got anything to confess?