Today I spent about 3.5 hours reading about various parts of GPS in a PDF that is on the u-blox website.
I spent about 2.5 hours today becoming familiar with interfacing TTL devices with RS-232. It looks like all it takes is a MAX233 chip, which converts the voltage levels and inverts the signals for transmit and receive functions. I might have to do this in order to interface my Etrex to the multi-I/O board. Since the I/O board’s UARTs can handle RS-232 levels (diode protection), I may be able to get by without the conversion if I just use software to invert the binary coming from the GPS before converting it to ASCII, etc. The only problem here is that the I/O board is expecting a low start bit and a high stop bit. This will be messed up without the hardware inverter. Oh, well.
Yesterday I found a sort of tutorial on writing c programs that interface with hardware at http://www.learn-c.com. I spent about 4 hours looking through the first few parts. I think I’ll have a better idea of what the computer is actually doing this time than when I was programming for the guided missile project.
I ordered a 6-DOF IMU kit from Rotomotion and a Flashlite 386EX developer’s kit with an I/O board from JK microsystems. These two components will be the backbone of the control system. Right now I’m planning on using my Garmin Etrex as the GPS.
I found a 4 Hz GPS a few weeks ago. Here is the link. www.u-blox.com/products
I spent about 1.5 hours working on the project today. I stumbled on to a Kalman filter paper that seems interesting. After talking to Sam, I think I’ll start out using one of the Flashlite computers to do the bulk of the onboard calculations and to handle the GPS and any other sensors. If I need more computational power later, I can add a gumstix computer and interface via serial. I will email JK Microsystems to see what are the important differences are between a 186 and a 386 Flashlite. At this point I’m going to concentrate on getting the autopilot hardware put together. I’ll choose what kind of airplane I want and get it built over Christmas break.
Yesterday I spent another 2 hours looking for small computers to run the flight controls on. I got a reply from Rotomotion. They indicated the configuration I’m planning to use is basically what they have been using on their machines. Now the basic question is how much computing power I’ll need to make the system work. I’m partial to the Flashlite computers, but I may have to move to a an X-scale board or something if 33 MHz isn’t fast enough.
Today I spent about 4 hours looking for servo control ideas and cheap high-speed computers to do the kalman filter stuff I’m anticipating for the control system. I found one good lead: Gumstix has a tiny 400 Mhz computer which runs linux. According to what I read on the autopilot.sf.net pages, anything over like 150 Mhz should be able to run all of the control and I/O routines. It seems that much computing time is devoted to the kalman filters.
I haven’t looked yet, but I think I will use a couple of pressure transducers for airspeed and altitude inputs. I think an angle of attack vane would be useful for estimating the flight condition. In the future, a sonar range finder might be of use for AGL altitude calculations for flight low to the ground.
It’s The Cheat!
Today I spent another 2 or 3 hours looking at possible hardware for the UAV.
I emailed Rotomotion today to see what they think about using their 6-DOF IMU as a sort of sensor hub in my control design. Right now I’m thinking of using either the flashlite 186 or the flashlite 386EX as the main onboard control computer. I think this will give a good amount of flexibility as far as the ability to add more sensory inputs and log data.
Nothern Tool has a reconditioned string trimmer with a 31cc engine for $69. This might be a good engine for the UAV.
I also looked into the possibility of using my FMA infrared wing leveler as a backup to the flight control system. It could also be used as a sensor input to the computer, just to have some more information to play with.