Making something innovative is always a challenge, while thinking of all of the possible project; Quadrocopters, Octocoptors, and other multi-rotors, we discovered that we wanted to have an idea that would help solve a problem. There’re many QuadCopter that we can buy with Gyro, self-balance and even GPS that supposedly would make the drone go wherever you want it go, but few people had ever done for a fixed wing. So we decided to do it, we can make it deliver literally everything that is under its capacity. Overall saying, a fixed-wing-plane is better than a conventional multi-rotors-copters in many ways. 

Deciding how our plane should look was quite easy, there are many available models that we can search up on the internet, you can see that from one of our early sketches. However unlike the usual design procedure, we first focused on the flight controls first. We choose Arduino because all of our group members are capable of programming in that language. However when we were programming on the Uno board, we encountered a problem, the Receiver doesn’t output normal dc power, it doesn’t even output usual PWM (which these kinds of controllers always do). 

It outputs a Pulse Train, though it works like a PWM signal, but there is a slight difference between them. They are all controlled by calculating the duration of the pulse, but in PWM signal, the gap between each pulse is strictly controlled, in the Pulse-Train, the pulse itself still matters but the gap doesn’t interfere with the signal as long as it is longer than 20 million seconds. In order to solve that problem, we downloaded a new library which allow the Arduino Uno board to understand the signal, it took us about half a week to complete all of the manual control programme. 

Then we started working on the plane, we used a one metre Carbon-Fiber rod for our ‘Frame’. Our first concern was how to connect the motor firmly onto the Carbon-Fiber Rod, we thought about simply screwing the motor onto the rod, but despite Carbon-Fiber’s strength, it is extremely easy to crack under horizontal force. To solve that, we created a 3D model in SolidWorks to connect the motor to the rod. When it finished printing, we installed the motor to the connector and push the motor on the rod, though it looked very tough and unmovable from the first look, when we attempted to try the motor with the prop, the motor and the connector both flew off the rod. Luckily no one was hurt, but we were all scared plenty. Finally we drilled a hole on the connector and drilled a hole onto the rod using the table drill learning our mistake from previous attempts with a hand-drill. We secured the motor and moved on to the next component. 

The wings came next, first we thought designing the wings would be a easy task: ‘What’s so hard to draw two rectangles?’, but as it turns out that we spent one day to design the wing in FLXR5 in order to have the most efficient design. Then we need the hot wire cutter to cut the wing out —— believe me, it’s harder than you would think. Because the foam boards are either used, bent, or too narrow. When we finally cut the wings out, we discovered that it was in the wrong scale. It took us about half a day to cut the wings out. Then comes the million-dollar-question: how are we suppose to connect two 20cm-wide, 75cm-long wings onto a one centimeter square, perfectly smooth rod?! The first thing we thought about was a connector, so we made one, but the ‘connector’s’ sole usage was to widen the rod’s width from one centimetre to six centimetres, so the question remains —— how are we going to connect the wings to the ‘connector’? We sure can’t screw it on. Then we thought about glue, normal super-glue sure won’t do, we need some kind of glue that would be act like a gel, so we put hot-glue on the connector first, then put the wings on it to prevent it from moving momentarily, then we mixed e-poxy and put it into the gap between the gap between the connector and wings so the wings would never fall off or move, we left it over night to dry. We did he same thing to the rudder and the tail. Then comes installing the control surfaces, the servos that should move the ailerons, rudder, elevator. This part was surprisingly easy, it only took us half a day finish all of the control surfaces. until they fell off

    On the second week, we started building the auto-nav system, we were able to get a Arduino Mega with the GPS, gyro and a XBee which is a ling-range transmitter. We had some problem Serial printing the numbers and the GPS was unable to connect to satellites indoor. However we gradually overcame them, but the problem was that the auto-control wasn’t stable enough for the plane to fly in the air, so we thought about PID algorithm. PID —— Proportion, integration and differential. It was the easiest to stabilize a programmable moving object, it prevent the plane from making a very sudden movement but yet has enough flexibility to maneuver the plane to do sudden rises and descents. Then a problem ran into our faces: the Uno board didn’t have enough ports and we’ll have to switch to Mega Board, but the PulseIn library only worked for the Uno board and will crash the Mega, so we asked the library provider to rewrite the library for us that we could use it for the Mega 2560 board. We spent about half a day hooking up the Mega board to the control surfaces and cleaning up the wires using zip-ties. However when we test flighted it, the elevator got blew off, the wings weren’t damaged but the gears broke so we had to laser-cut it again. We had a little time to adjust our KP value and get the PID algorithm to work.

    In general our project went very well, our fixed-wing can fly over 40 miles per hour, reaching a range of more than 40 miles and carrying a cargo of at least half a kilogram. We can use it to drop medicine, inflatable life preserver, water purifier and even more. It can switch between manual control and autonomous flight. Our plane should be able to fly fully without human interfere though we had never have a chance to test it.