The deadly stink of progress
So. Things are ticking along at the moment but its not so exciting. I had a meeting with my supervisor and he gave me some much needed criticism to get my work onto the right track for my final year. However, it winded me right before I needed to be getting my presentation sorted for the upcoming TAROS conference. I had been working really hard on my next paper, which would of made up the future work section of my presentation, but Alan told me it wasn't worth investigating. It was a bit of a kick in the nuts. He also said that he expected me to be futher along at this point in my PhD, that my current vein of research isn't credible for entry into my thesis. I think these are all good things to hear, in an "omg, is that smell methane? we better get out of this coal mine" way.
My colleague behind me was receiving advice on writing his thesis, and he only stared a few months before me. I feel that because he is involved in a big research project his PhD has been so much more directed and easy. But thats the grass is greener, i suspect. He hasn't had as many holidays abroad as I have :) Nor has he been involved in so many side projects. And he doesnt get to choose something he finds interesting either. Swings and round-abouts.
But the good news is that I am starting to stop pitying myself so much and getting on with things again. I just need to think really hard at the moment and I have forgotten what that is like. I've been working on my presentation and it is forcing me to reassess my research motivations and recall earlier ideas. I've had the same fundamental problem with my PhD the whole way through, and I just forgot (or side-stepped it). After Alans criticism I felt like he had dumped an even bigger problem in my lap, but really, its the same old demon. I need to set my focus back to solving this big problem, even subconsciously outside of working hours.
I've made some nice progress programming wise, but this is a dangerous trap for me. I get huge amounts of satisfaction from solving programming problems, and there is always something to fix-up or solve in the PhD, or even nice little ways to solve miscellaneous problems when I am bored. Often, it helps to reveal the way forwards for me. But more often, I tinker tinker tinker. Alan suggested that I 'cheat' and use the motion tracking system to allow the robots to have information like the position of their neighbours and perhaps their location with respect to virtual areas of the arena. He knows I have some form of aversion to providing abstract and virtual information for the robots. However, I think this has changed in me now. I always maintained that stance to keep my work credible. But the truth is, the robots I use would not function outside of the lab, so who cares if I give them a virtual sensor of somekind?
So last week, after Alan suggested I should create 'virtual patches' and that it might take a week or so to implement, I fully implemented it in 3 hours, from scratch. Smug with myself, I was thinking it wasn't so hard. But actually, I integrated the motion tracking system, the variable number of robots, and a proxy for the two that spans 3 network domains. I put some nice features in too; you can specify any number of 'patches' into a text file, which is automatically read and implemented. The patches can also support a value of some kind, which I supposed could represent the fertility of a food source, or the remaining food items, or even signify the importance of the location (nest, goal, etc). It works over WiFi, so again some fun with TCP/IP sockets, and I wrote the client to run as a parallel thread on the robot. This is the most crazy system I have ever built/coded, but I've had two years of tinkering to do it in.
But yeah. None of that is really important to my research. However, my simple stupid robots now have more information at their disposal. I just have to figure out a nice problem for the robots, and how to specify a utility metric to evolve behaviour to solve the problem in a cooperative, self-organised way. I read a great paper on robots that explore, navigate and divide labour to forage for food (inspired by mould); which is all within the powers of my robots. They simulated it, I can implement it. However, I need to match the scenario to my algorithm, and find a way to make a significant advancement. I could evolve the behaviour (probably), but Alan says my algorithm must take information from the real world to improve the performance in some measurable way, otherwise using real robots is insignificant. It is pointless me evolving the real morphology of the robot, as the robots are simple and the differences between robots are trivial. It is far more useful for me to measure environmental or social cues in the swarm, as these are the basis of the interactions that form group behaviour.
Well, this is what is going around in my head. I'm feeling a bit sick of always carrying this shit around. And a bit bored of the elusive light at the end of the tunnel. A bit bored of if I am doing enough, or doing it right, or wondering how it will all turn out. One way or the other, I have 12 months left.
My mum and dad are going on holiday to visit my uncle in france. I am so pleased for them, I have never known them to go on holiday together. I know they went to San Francisco 30 years ago or so. They are even making it a holiday-holiday, going to Bordeaux via Spain! It is quite a surprise for me, and I guess it marks the point at which my parents are happy with their childrens progress? We are self-sufficient, trustworthy individuals and my parents are now free to live unburdened.
My colleague behind me was receiving advice on writing his thesis, and he only stared a few months before me. I feel that because he is involved in a big research project his PhD has been so much more directed and easy. But thats the grass is greener, i suspect. He hasn't had as many holidays abroad as I have :) Nor has he been involved in so many side projects. And he doesnt get to choose something he finds interesting either. Swings and round-abouts.
But the good news is that I am starting to stop pitying myself so much and getting on with things again. I just need to think really hard at the moment and I have forgotten what that is like. I've been working on my presentation and it is forcing me to reassess my research motivations and recall earlier ideas. I've had the same fundamental problem with my PhD the whole way through, and I just forgot (or side-stepped it). After Alans criticism I felt like he had dumped an even bigger problem in my lap, but really, its the same old demon. I need to set my focus back to solving this big problem, even subconsciously outside of working hours.
I've made some nice progress programming wise, but this is a dangerous trap for me. I get huge amounts of satisfaction from solving programming problems, and there is always something to fix-up or solve in the PhD, or even nice little ways to solve miscellaneous problems when I am bored. Often, it helps to reveal the way forwards for me. But more often, I tinker tinker tinker. Alan suggested that I 'cheat' and use the motion tracking system to allow the robots to have information like the position of their neighbours and perhaps their location with respect to virtual areas of the arena. He knows I have some form of aversion to providing abstract and virtual information for the robots. However, I think this has changed in me now. I always maintained that stance to keep my work credible. But the truth is, the robots I use would not function outside of the lab, so who cares if I give them a virtual sensor of somekind?
So last week, after Alan suggested I should create 'virtual patches' and that it might take a week or so to implement, I fully implemented it in 3 hours, from scratch. Smug with myself, I was thinking it wasn't so hard. But actually, I integrated the motion tracking system, the variable number of robots, and a proxy for the two that spans 3 network domains. I put some nice features in too; you can specify any number of 'patches' into a text file, which is automatically read and implemented. The patches can also support a value of some kind, which I supposed could represent the fertility of a food source, or the remaining food items, or even signify the importance of the location (nest, goal, etc). It works over WiFi, so again some fun with TCP/IP sockets, and I wrote the client to run as a parallel thread on the robot. This is the most crazy system I have ever built/coded, but I've had two years of tinkering to do it in.
But yeah. None of that is really important to my research. However, my simple stupid robots now have more information at their disposal. I just have to figure out a nice problem for the robots, and how to specify a utility metric to evolve behaviour to solve the problem in a cooperative, self-organised way. I read a great paper on robots that explore, navigate and divide labour to forage for food (inspired by mould); which is all within the powers of my robots. They simulated it, I can implement it. However, I need to match the scenario to my algorithm, and find a way to make a significant advancement. I could evolve the behaviour (probably), but Alan says my algorithm must take information from the real world to improve the performance in some measurable way, otherwise using real robots is insignificant. It is pointless me evolving the real morphology of the robot, as the robots are simple and the differences between robots are trivial. It is far more useful for me to measure environmental or social cues in the swarm, as these are the basis of the interactions that form group behaviour.
Well, this is what is going around in my head. I'm feeling a bit sick of always carrying this shit around. And a bit bored of the elusive light at the end of the tunnel. A bit bored of if I am doing enough, or doing it right, or wondering how it will all turn out. One way or the other, I have 12 months left.
My mum and dad are going on holiday to visit my uncle in france. I am so pleased for them, I have never known them to go on holiday together. I know they went to San Francisco 30 years ago or so. They are even making it a holiday-holiday, going to Bordeaux via Spain! It is quite a surprise for me, and I guess it marks the point at which my parents are happy with their childrens progress? We are self-sufficient, trustworthy individuals and my parents are now free to live unburdened.