The thermistor calculations which I posted seems to be getting quite a large viewing. I though that once I stopped posting, there would end up next to no page views but instead there has actually been a steady flow, mostly for the thermistor calculations.
I've also just applied for youth allowance, so I should in the next month or so have enough money to buy a RepRap Pro. I'm not sure whether to buy a tri-colour or mono yet though.
 |
| Mendel Mono from RepRap Pro |
I think my first upgrade will be to make the system ARM based rather than Audrino (Atmel AVR) based. This is so I can learn ARM programming as well as set up networking and a display with controls for the printer too. The RepRap site has links to three projects, 4pi, SmoothieBoard and R2C2. R2C2 seems to be the most focused on 3D printing as well as being the most feature rich (from my brief research). These all have code for G-code processing so I can base my initial firmware on these.
Just before Easter, I ordered and received Bloodhound SSC's book "The Science Behind the Speed". Its a really good book that tells the story of the problems the project had to over come to even consider breaking the land speed record and aiming for 1000mph. If anyone hasn't looked into the project, its quite interesting and would recommend, even if your not into motorsport as its much more about education of science, engineering and technology than it is about the speed record. The part of that I most enjoyed was the aerodynamics and how its so vital to the project.
After I finished the Bloodhound book, I started back reading my artificial intelligence book. I'm in a section where its all about deterministic logic and its quite hard to read. So because of this, I got side tracked and tried to find a book on what I thought was called visual intelligence but seems to be called computer vision. I eventually found a few and started reading about how images can be manipulated (I haven't got very far through the book yet). This led my mind astray again, to drones.
For quite some time, I've wanted a multi-copter drone. I had no idea what I would use it for but due to the nature that it is very mechatronics based always loved the idea. I've now thought of an idea, that is to test out visual intelligence or computer vision techniques on one so that it can autonomously follow an object and avoid objects while hopefully trying to minimise other sensors. Other projects that I have read about seem to always pile in a mass of other sensors. This is because I want it to be much more like nature if that makes sense, ie humans etc don't have laser sensors etc. (I sort of believe that one reason for going into robotics and artificial intelligence is to gain a better understanding of us as human beings (and nature in general) while also then trying to create a copy of this to show that we do actually understand it not just think we do). This drone, UAV or UAS what ever you feel like calling it, could then be used as a camera for movies, surveillance, robotic assistant or just develop technology for avoiding obstacles.
I would be quite happy to buy a pre built drone but the ones I could find online are all several thousand dollars or are toys such as the Parrot Ar Drone. The Parrot Ar Drone does have one feature that I do quite like in that it has safety guards round the props where it seems that most other models don't have this feature. I wouldn't make them quite as thick though, or make them from foam as the Parrot Ar Drone does.
Another reason to build my own is to experiment with power sources. I think flight time is obviously a key issue and so using the pre-built models they seem to only have 20min to 30min max flying time. I would love to employ some renewable energies such as solar or even going quite ambitious and using fuel cells. Another way to make them more power efficient, is to make the rotors/props larger (or simply more, I'm thinking six or eight). You use more energy on the whole but the payload capacity goes up at a greater rate. Since realising this, I've thought I could potentially make a larger scale one and have it almost a meter in diameter. This would all help with the ability of being able to attach a second camera such as a DSLR or heavier sensors for the autonomous system.
The 3D printer would help with this for making parts for the core but also to make moulds that can then be used to make carbon fibre versions of supports. The reason for using carbon fibre, is to that the copter is lighter and thus uses less power to fly so more power can be used on the intelligence processing. I'm also going to try and learn some basic aerodynamics and apply it to the structure as I was inspired by the Bloodhound SSC project.
Once built, and before I start on the autonomous systems, I'm thinking that I will have to port over one of the few auto-pilot software kits. Hardware wise, I'm hoping that I'll be able to use the project to understand vibrations and damping which I can apply to camera stability. You can see from this footage of the recent Dunalley Fires, Tasmania, how the vibrations are affecting the footage. This could be fixed digitally but would be energy efficient to use mechanical damping as much as possible.
I think the drone should be a nice project to start off my adventures into home engineering projects and should be a nice way to make something fun but also functional and educational.
