Heh, learning about microcontrollers isn't the purpose of the project, but an added side effect. I've learnt a lot about microcontrollers and their internal structure during some of my subjects, so hopefully I'll be able to make the code more efficient. The ardupilot is one example of a plane UAV based on the Arduino. You might be able to get a little helicopter to modify and make fly around your room, but it means it won't be able to lift a lot of weight.
Some arduino projects try to use a Kalman filter, however because of the time taken to process them, they can't do large Kalman filters. The current implementation that we have currently, it takes 1012uS on the Arduino to run the function, but only 56uS on the Maple. However it's only a single state Kalman filter that my project partner found on the internet to try out, he's designing the Kalman filter and hopefully will inform me to how they work (mathematically), since they're not one of my strong points. We're using the current implementation to learn off so we can hopefully expand on it. I can't stay I've looked at st.com for a library on kalman filters as from my understanding, it requires a dynamics model of the system to be able to predict where the plane will be next in 'x' time.
The switch to the Maple Mini is for 2 reasons, making the electronics smaller and preventing my current Maple (it's my special Maple since my girlfriend got it for me) from being damaged in the event of an accident. I believe that the Maple Mini will have both enough IO and memory for the project, but I'm not entirely sure on the processing power. Some of the tasks I'm planning might be offloaded onto a CPLD or FPGA. The Maple currently is just going to be serving as a communication hub as we work out the kalman filter, and will be feeding values over a set of XBee modules to probably MATLAB.
We will have a few systems to prevent the Aircraft from causing injury to anyone. We're going to be implementing a heartbeat signal (if it doesn't get the signal in 'x' time, it will shut down), a emergency shutdown signal (if a certain packet is received, it shuts off everything) as we'll be getting telemetry data from the robot, and for initial testing, a harness. We have a few spare parts however, we wish to prevent a crash. The university gives us a 'small' amount of funding for our project, but it's the time to get new parts that will cause problems.
Eventually we're planning to have a GPS module on the robot and for it to navigate through waypoints, the initial test is just getting the robot to hover stably. We need to work out where to test as we did a light test in the robotics lab, and had some valid complaints from other students using the lab. Propellers are fairly cheap, and we have another set if needed thanks to another group member from last year doing the same project (and copying all of the information from the first semester we did it together). We bought the brushless motors and initial components from http://www.rcheadquarters.com.au/ but have gotten replacement components from http://www.hobbyking.com/ due to the components being a lot cheaper, although they don't have the same motors we have. If someone burns out, I'll probably go back to rcheadquarters if we can't wait. I've had to in the first semester after we accidentally blew up a brushless DC speed controller. The requirements were for the engines to be able to stay under 25A and be able to spin the propellers fast enough to generate the lift required, at full power, they consume around 280W each. We're using a 5Ah 25C 3S1P Lithium polymer battery to power them and the electronics. In tests, the BEC (Battery Elimination circuit, just a linear voltage regulator to remove the receiver battery pack) on the speed controller feeds the Maple 5.6v via the Vin pin for power.
I will keep this thread posted on the progress of the project, as for posting it in the projects section of the Wiki, I'll have to see, if the leaflabs team wants me to, then I probably will.
