Table of Contents
Project Overview / Major Emphasis 2
Chronological Review 2
Problems Faced 14
Lessons Learned 14
Functional Specification Document 15
Concept Generation and Selection 18
Project Overview / Major Emphasis
The lab is currently using a hexakopter to implement the different software and hardware ideas that are being designed for the eventual competition. The hexakopter currently needs some modifications and additions to help facilitate this process. Our group’s responsibility is for the structural implementation of the overall team’s needs. With that background, our group has three short term goals, including: 1) Designing a hard mount for the Bumblebee camera that allows the camera to face either forwards or down. 2) Redesigning the computer mount that is more accessible that includes a protective computer guard. 3) Installing rotor guards to protect against prop strikes. Then, after creating a working model with the hexakopter, we will then work on creating a light weight platform that conforms to the weight requirements. This will most likely be implemented with a quad-rotor craft.
Chronological Review
January 26, 2011
We divided up into teams. James, Mike, and Larry are on the “Building Group” team.
January 28, 2011
We broke out into our groups today. We received our expectations and timeline information.
The three short term tasks, due February 18 are
1- Redesign the computer mount and include a computer guard.
2- Create rotor guards.
3- Create a camera mount that can look either forward or down.
Larry took responsibility for task 1, the computer mount. James took task 2, create rotor guards. Mike took task 3, create a camera mount.
The long term task is to prepare a working lightweight platform weighing less than 1.5 kilos that can carry all the necessary hardware. This is an end of semester goal.
January 31, 2011
Began gathering materials for the prop guard on the HexaKopter:
3 - .12”X36 carbon fiber rods
6 - Rubber T connectors (goodwinds.com)
3 – Aluminum connecting tubes to create the carbon fiber loop ID .135”, OD .19”, Length 2”
Zip ties
VHB double sided tape
February 2, 2011
James obtained a carbon fiber strip .07”X.433”X36” to build supports then cut it into 6 5.5” lengths. After cutting the supports to length, James mounted the prop guard to the functional HexaKopter.
Mike worked on and oriented himself to the HexaKopter CAD files. He sketched some of the files out on paper and wrote down the measurements of the camera, and started sketching some ideas out. He took some measurements of the actual HexaKopter to start refining the CAD file. The main goal in doing this is to have the holes that will be available in the base that I can use to fit the mount to. If time is lacking, we may just get the specs on the holes we want to use, and add those to the CAD file.
February 4, 2011
Due to time constraints, Mike decided not to update the CAD file and just create the mount parts in CATIA using holes from the HexaKopter.
February 7, 2011
Mike went to the CB shop, and they had some steel pieces that were already bent. This saved a lot of time having to get a perfect 90 degree angle out of the aluminum. After cutting them to size, he manually drilled the holes in the machine shop.
February 9, 2011
We got the Helicoils ordered from McMasters.
They have the following properties:
Stainless Steel Type 18-8 Stainless Steel
Finish/Coating Plain
Threaded Type Right-Hand Threaded
System of Measurement Metric
Internal Thread Size M3
Internal Metric Pitch .5 mm
Insert Length 4.5 mm
Length Note Insert Length given refers to the installed length.
Maximum Outside Diameter 4.35 mm
Drill Size 3.2 mm
Rockwell Hardness C42 - C50
Minimum Tensile Strength 200,000 psi
February 14, 2011
Drilled the holes in the camera for the helicoils, tapped them and installed the inserts.
February 18, 2011
Here are picture of the completed short term tasks:
1: Computer Mount
2: Rotor Guards
3: Camera Mount
February 21, 2011
Mike and Larry used a bifilar pendulum to calculate the moments of inertia of the HexaKopter with the computer installed and a dummy weight to simulate the camera.
February 22, 2011
Larry purchased the L4 Power Board from quadrocopter.com today. We needed a way to wire up the speed controls in an efficient way. The HexaKopter also has a power distribution board that works very well, so I decided to find the equivalent one for a quadrotor. There were a couple of variations, but this one will fit under the Pelican frame.
March 2, 2011
Larry worked with Everett on soldering the new power distribution board. We decided to omit the power switch in order to conserve weight. This wires all the speed controls together to eliminate wire clutter and extra weight.
March 9, 2011
We worked on selecting the motors for quadrotor. We have two available motors. On is the same motor that we are using on the HexaKopter. It is the 2827-35 motor. Another one that we have available is the MK3538 motor. It is a heavy duty motor. We compared the motors and selected the set that would weigh the least and do the job. We selected the 2827-35 based on the results from the comparison.
Motor / ROXXY2827-35 / MK3538Weight (g) / 57 / 125
Recommended Current / 4 to 9 Amps / max 20 amps
current for 400 g thrust / 3 / 2.5
Total Current Draw / 12 / 10
Battery for 10 min flight mAh / 2222 / 1852
Combined weight of motors / 228 / 500
Battery Weight (g) / 238 / 192
Total Weight / 466 / 692
March 16, 2011
We worked with Rob today and talked about system ID. Rob taught us how to use the cortex room and fly the HexaKopter. We also talked about the system ID toolbox in Matlab. We are going to try and replicate the pitch system ID on the HexaKopter that Caleb performed in his thesis. Once we accomplish this we are going to work with the graduate students to do system ID on the quadrotor and design a controller for the quadrotor.
March 17, 2011
Larry finished soldering up the power distribution board and the flight controller for the quadrotor.
March 18, 2011
We finished building the quadrotor today. James put on the rotor guard and Larry wired up the Spektrum receiver to the flight control board. We flew the quadrotor with RC command. It seems very stable and fightable. However, when testing the yaw control, the quadrotor suddenly went out of control and performed a full 360 degree barrel roll and landed on its landing gear. A propeller was broken in the process but no other damage occurred. This same type of crash has happened once before with the HexaKopter. The exact cause of the crash is unknown but we suspect that it has something to do with some glitch or instability in the onboard flight control board.
In addition, we flew the HexaKopter in the cortex room and collected system ID information. We put the HexaKopter through a variety of exercises to test the pitch control. Using the already in place LQR controller, we commanded position inputs of a sinusoidal nature and recorded the commanded pitch and the actual pitch response. We tried several sum of signs inputs as well as a swept sign input. We hope that we covered a sufficient frequency range to get a good system ID. Mike wrote a Matlab program to parse the data.
% undergrad_extract.m
% extract columns of desired data from given data log
% change format to long
format('long');
% read data from file
% load log.txt, spaced columns, 1 row heading;
log = importdata('log.txt', ' ', 1);
% copy columns into files
time = log.data(:,1);
theta = log.data(:,6);
command_pitch = log.data(:,9);
% create n x 3 matrix with time, theta, command_pitch
mat = [time,theta,command_pitch]
% save to tab-delimited 16-digit ASCII format
% FYI: to save to space-deliminated instead of
% tab-deliminated, delete -tabs
% FYI: format is 'save file.name variable formats'
save log_data.txt mat -ascii -double -tabs;
%plot data
figure(1)
subplot(2,1,1);
plot(time,theta);
subplot(2,1,2);
plot(time,command_pitch);
March 19, 2011
Larry worked on getting an estimate for the total weight of the quadrotor and sensors.
Weights / oz / gFrame / 6 / 170
Flight Control / 0.9 / 26
ESC board / 1.4 / 40
4x Motor & Screws / 11.2 / 318
Battery / 8.4 / 238
Computer / 6 / 170
Rotor guard / 3 / 85
Laser / 5.1 / 145
Misc / 3 / 85
Total / 45 / 1276
Remaining / 7.8 / 224
March 21, 2011
We performed the system ID analysis on the previously collected data using the Matlab SysID toolbox. We tried several different models and it appears that the ARX441 fits the data the best. Next we plan to use the ARX441 transfer function to develop a position controller for the HexaKopter.
ARX441 Model
Discrete-time IDPOLY model: A(q)y(t) = B(q)u(t) + e(t)
A(q) = 1 - 0.593 (+-0.03288) q^-1 - 0.1743 (+-0.03819) q^-2
- 0.0895 (+-0.03813) q^-3 - 0.07465 (+-0.03173) q^-4
B(q) = -0.0003355 (+-0.0001852) q^-1 - 0.0005413 (+-0.0001623) q^-2
- 0.0004597 (+-0.0001612) q^-3 - 9.198e-005 (+-0.000165) q^-4
March 23, 2011
We worked on making a Simulink block diagram to simulate the response of the HexaKopter. We ran into problems getting the preexisting pitch commands to feed into our Simulink model. After talking with some of the graduate students in the lab, we decided that we needed to write a function that would pass in the commanded pitch values for each time step.
March 25, 2011
We had our 2nd design review today.
March 28, 2011
We tried flying the quadrotor again today, but the flipping problem persists. It seems that within the first minute or two of flight, the quadrotor will perform a hard roll to the right and then crash. We started to investigate the cause of the flipping. We hooked up the KopterTool to the quadrotor flight control board. It is a communication device and program that allows us to see all the sensor values that the flight control board is seeing. At first we powered up the quadrotor and just wiggled it around while watching the sensor readouts. We saw the following sensor spike on the pitch angle.
We then held the quadrotor to the ground and powered up the motors and watched the sensor readouts. This time the pitch accelerometer value spiked.
March 31, 2011
Larry created an account on both the international and the US based MikroKopter forums and made a post about the quadrotor flip. There were a couple of responses, mostly stating that we probably needed to resolder the accelerometer sensors.
April 4, 2011
We are having trouble figuring out exactly what is wrong with the flight control board. We decided to take the working board off of the HexaKopter and put it on the quadrotor. Everything worked well and the quadrotor flew fine. It seems that the problem is indeed a faulty flight control board. We are thinking that since the board was defective from the very beginning, we should be able to get a replacement one.
Problems Faced
It really was unfortunate the quadrotor flight control board turned out bad. It really hindered our progress in developing a controller for the quadrotor. We spent a lot of time trying to figure out what was wrong with the board instead of doing research.
Due to the unexpected flipping of the quadrotor, the rotor guards broke off. They were subjected to unexpected loads that they were not designed for.
A lack of clear direction at the beginning of the project got us off to a slow start. The first several weeks of the project were unproductive in accomplishing what turned out to be our final project goals.
Lessons Learned
If we could go back and change one thing, we would have built the quadrotor sooner so that we would have had more time to troubleshoot it and hopefully have gotten a controller working for it.
A clear project definition helps us to work efficiently. We should have been more proactive in filling in the questions on the lightweight platform requirements.
Functional Specification Document
Introduction
This document has been prepared to help our group, the team, and our customer (Dr. Beard - and the different competitions this work will assist in completing) understand the basic concepts we plan on focusing on this semester. It lays out the ideas we plan on implementing, and the ways we are going to measure that implementation. In the end, it gives a preliminary idea of what we will be developing over the next couple of months.
Background
The lab is currently using a hexakopter to implement the different software and hardware ideas that are being designed for the eventual competition. The hexakopter currently needs some modifications and additions to help facilitate this process. Our group’s responsibility is for the structural implementation of the overall team’s needs.
Project Description
With that background, our group has three short term goals, including: 1) Designing a hard mount for the Bumblebee camera that allows the camera to face either forwards or down. 2) Redesigning the computer mount that is more accessible that includes a protective computer guard. 3) Installing rotor guards to protect against prop strikes. Then, after creating a working model with the hexakopter, we will then work on creating a light weight platform that conforms to the weight requirements. This will most likely be implemented with a quad-rotor craft.
Project Requirements
We have been hired on as contractors to work on some specific tasks in parallel with the graduate students. The graduate students are effectively our customers. They have given us some tasks that they would like completed.
The following is a table of requirements as stated by our customers.