39369044 Internet Mobile Robot the Quad Copter

Internet Mobile RobotThe QuadcopterGroup 13 Rahul Bura Mohamed Chande Vinayak Goge Hue Vo

Supervisor: Professor Peter X. Liu

A report submitted in partial fulfillment of the requirements of SYSC-4907 Engineering Project

Department of Systems and Computer Engineering Faculty of Engineering Carleton University

April 7, 2010

Internet Mobile Robot

April 7th, 2010

AbstractGroup 13 consisting of Rahul Bura, Mohamed Chande, Vinayak Goge and Hue Vo have developed a start-up project as a partial fulfillment for SYSC 4907 Engineering Project: Internet Mobile Robot: The Quadcopter. The objective of the project was to design and implement a Quadcopter (helicopter with four propellers) that can take flight and be controlled using a remote client application over the internet. Currently, different Quadcopter designs have been implemented. However, most of them have used handheld Radio Control implementations. With the design implemented in this project, different applications can be developed to control the Quadcopter over the internet from a remote location. This opens up different possibilities with the design being applied in different areas ranging from Surveillance to Virtual Gaming Technologies to Military Applications. Different designs were explored and from these designs, it was determined that we would need a microcontroller to run control algorithms, a Wi-Fi chip to facilitate wireless communication and a webpage to for user interaction. Any command received from the WI-Fi chip is processed by the microcontroller and executed by all the components including motors, speed controllers and inertial measurement units that are used to stabilize the Quadcopter. The communication system implemented a server/client architecture with the Wi-Fi chip behaving as a server that acts to reply to requests from a remote client webpage. For the data transfer mechanism, TCP protocol was used over the Internet to send traffic from the client to the server and bidirectionally. The webpage, implementedDepartment of Systems and Computer Engineering |Carleton University i


April 7th, 2010

as a simple Graphical User Interface (GUI), is a good replacement for the RC handheld devices and allows for easy portability from one client to another.

Department of Systems and Computer Engineering |Carleton University

ii


April 7th, 2010

AcknowledgementsWe would like to formally thank Professor Peter X. Liu for supervising and for his guidance throughout the Internet Mobile Robot project as well as giving us the opportunity to explore and experiment with our creativity. We would also like to thank the Technical Support Staff, Danny Lemay, Jerry Buburuz and Daren Russ, for their support throughout the project. Their prompt actions, experience and knowledge allowed us to surpass various milestones throughout the duration of the project.


iii


April 7th, 2010

Table of ContentsAbstract...................................................................................................................................... i Acknowledgements ................................................................................................................. iii List of Figures ........................................................................................................................... vi List of Tables .......................................................................................................................... viii 1.0 Introduction ...................................................................................................................... 1 1.1 Background .................................................................................................................. 1 1.2 Motivation ................................................................................................................... 3 1.3 Problem Statement...................................................................................................... 3 1.4 Proposed Solution and Accomplishments ................................................................... 3 1.5 Overview of the Remainder of the Report .................................................................. 6 2.0 The Engineering Project .................................................................................................... 8 2.1 Health and Safety......................................................................................................... 8 2.2 Engineering Professionalism ........................................................................................ 9 2.3 Project Management ................................................................................................. 10 2.4 Individual Contributions ............................................................................................ 11 2.4.1 Project Contributions ......................................................................................... 11 2.4.2 Report Contributions ......................................................................................... 12 3.0 Robot Design ................................................................................................................... 13 3.1 Structure .................................................................................................................... 13 3.1.1 Number of Motors ............................................................................................. 13 3.1.2 Frame ................................................................................................................. 15 3.2 Communications ........................................................................................................ 15 3.3 Flight and Stability ..................................................................................................... 16 3.4 Flight Control ............................................................................................................. 18 3.5 Power ......................................................................................................................... 20 4.0 Hardware Components and Construction ...................................................................... 22 4.1 Frame and Structure .................................................................................................. 22 4.2 Microcontroller .......................................................................................................... 24 4.3 Flight and Stability ..................................................................................................... 25 4.3.1 Propeller and Motor Combination Configuration.............................................. 25 4.3.2 Electronic Speed Controller (ESC) ...................................................................... 26 4.3.3 Six Degrees of Freedom (DOF) ........................................................................... 27 4.4 Communications ........................................................................................................ 28 4.5 Battery ....................................................................................................................... 28 4.5.1 Flight Time and Battery Power Dependancy ..................................................... 29 4.6 Design Implementation and Final Structure .............................................................. 30 5.0 Stability and Manoeuvre ................................................................................................. 33 5.1 Filtering Noise ............................................................................................................ 33 5.1.1 Noise Reduction ................................................................................................. 33 Department of Systems and Computer Engineering |Carleton University iv


April 7th, 2010

5.1.2 Second Order Complementary Filter ................................................................. 34 5.2 Feedback Control ....................................................................................................... 35 5.2.1 Six Degrees of Freedom (DOF) ........................................................................... 36 5.2.2 System Control Theory ....................................................................................... 37 5.2.3 Proportional, Integral, Derivative Controller ..................................................... 40 5.2.4 Feedback Control Mechanism: Inertial Measurement Units ............................. 42 5.2.5 Feedback Control Loop ...................................................................................... 43 5.2.6 Flight Tuning Using Ziegler-Nichols Rules .......................................................... 45 5.3 Flight Configuration and Simulation .......................................................................... 46 5.3.1 Flight Configuration Methods and Tools............................................................ 46 5.3.2 Pre-Flight Tests................................................................................................... 49 5.3.3 Flight Control and Results .................................................................................. 50 6.0 Wireless Communication (Server) .................................................................................. 54 6.1 Communication System Overview ............................................................................. 54 6.2 Wireless Standards .................................................................................................... 55 6.2.1 ZigBee ................................................................................................................. 56 6.2.2 Wi-Fi ................................................................................................................... 57 6.3 WiShield Configurations ............................................................................................ 58 6.3.1 Network Type ..................................................................................................... 58 6.3.2 TCP vs UDP ......................................................................................................... 59 6.4 WiShield Functionality ............................................................................................... 64 6.5 Serial Peripheral Interface ......................................................................................... 66 6.6 Challenges and Solutions ........................................................................................... 69 6.6.1 Debugging .......................................................................................................... 69 6.6.2 Pin Conflict ......................................................................................................... 71 7.0 Wireless Communication (Client) ................................................................................... 73 7.1 Client Design .............................................................................................................. 73 7.2 Client Process............................................................................................................. 77 8.0 User Interface ................................................................................................................. 79 8.1 Software Requirements ............................................................................................. 80 8.1.1 Operating System ............................................................................................... 81 8.1.2 XAMPP................................................................................................................ 81 8.2 Challenges and Solutions ........................................................................................... 83 9.0 Software Integration ....................................................................................................... 85 10.0 Production Expenses ....................................................................................................... 88 10.1 Material Costs ............................................................................................................ 88 11.0 Conclusion and Recommendations ................................................................................ 90 11.1 Conclusion.................................................................................................................. 90 11.2 Recommendations ..................................................................................................... 92 References .............................................................................................................................. 93 Department of Systems and Computer Engineering |Carleton University v


April 7th, 2010

List of FiguresFigure 1: Rotating pairs of rotors [3]................................................................................................ 1 Figure 2: First Generation Design of De Bothezat Quadrotor [3] .................................................... 2 Figure 3: Wound Rotor [10] ........................................................................................................... 16 Figure 4: Brushless Motor Design [11] ........................................................................................... 17 Figure 5: Typical Microcontroller ................................................................................................... 19 Figure 6: Sketch of Proposed Structure ......................................................................................... 23 Figure 7: Initial Frame .................................................................................................................... 23 Figure 8 : Aeroquad Shield ............................................................................................................. 28 Figure 9: Graph of Flight Time versus Battery Life......................................................................... 30 Figure 10: Wiring Diagram for IMR ................................................................................................ 31 Figure 11: Final Design ................................................................................................................... 32 Figure 12: Second Order Complementary Filter [15]..................................................................... 34 Figure 13: Second Order Complementary Filter [15]..................................................................... 35 Figure 14: Possible independent movements in 3D space [16]..................................................... 37 Figure 15: Types of Control Systems: (a) Open Loop (b) Feed-Forward (c) Closed Loop [18] ....... 38 Figure 16: How the damping constant affects the time it takes to reach steady state [19] ......... 40 Figure 17: PID Controller loop [20] ................................................................................................ 42 Figure 18: Feedback loop including system, controller and sensor configuration [21]................. 43 Figure 19: Feedback Control Loop with control components ....................................................... 44 Figure 20: Motor Command outputs (S) during simulation of flight with Serial Monitor ............. 46 Figure 21: Sensor Data output (Q) during simulation of flight with Serial Monitor ...................... 47 Figure 22: AeroQuad Configurator GUI with updatable flight parameters ................................... 48 Figure 23: Various sensor outputs with AeroQuad Configurator GUI ........................................... 49 Figure 24: Original Communication System Design ....................................................................... 55


vi


April 7th, 2010

Figure 25: TCP three-way handshake ............................................................................................ 60 Figure 26: Re-designed Communication System ........................................................................... 63 Figure 27: WiShield Functionality .................................................................................................. 64 Figure 28: Handling TCP connection .............................................................................................. 65 Figure 29: SPI bus, single-master single-slave ............................................................................... 67 Figure 30: WiShield Schematic [27] ............................................................................................... 68 Figure 31: Simple Compiler provided for WiShield ........................................................................ 71 Figure 32: Ideal Layout of Client System in relation with WiShield ............................................... 74 Figure 33: Actual Implementation of Client System in relation with Wi-Shield ............................ 75 Figure 34: Client Design ................................................................................................................. 76 Figure 35: Client Process ................................................................................................................ 78 Figure 36: User Interface ............................................................................................................... 79 Figure 37: Components of the User Interface ............................................................................... 80 Figure 38: Pin 5 of SM (with WiShield) connected to digital pin 32 of PM using a wire [39] ........ 87


vii


April 7th, 2010

List of TablesTable 1: Report Contributions........................................................................................................ 12 Table 2: Flight Times for 2 Different Designs ................................................................................. 14 Table 3: Matching Motors for EPP1045 Propeller ......................................................................... 25 Table 4: Total Weight Calculation .................................................................................................. 29 Table 5: PID gain using Ziegler-Nichols Tuning Rule [17] ............................................................... 45 Table 6: Values of KP and corresponding qualitative observations for Ziegler-Nichols Rule ......... 52 Table 7: Relevant differences between TCP and UDP ................................................................... 62 Table 8: List of Materials and Respective Cost .............................................................................. 89


viii


April 7th, 2010

1.0 IntroductionWith the recent developments in wireless communications technology and its increased accessibility and affordability, network-based applications are now able to expand into new areas. One of these areas is the domain of Internet Robots. Internet robots have numerous applications and they can be programmed to perform multiple functions. The Internet Mobile Robot (IMR) constructed is an Unmanned Aerial Vehicle (UAV) that will be controlled wirelessly via a web browser.

1.1

BackgroundThe Unmanned Aerial Vehicle built is also known as

a Quadcopter or Quadrotor owing to the fact that it has four motors and propellers that stabilize and manoeuvre the robot. Flight control is achieved by varying the speed of each rotor to change the thrust and torque about the center of rotation. The rotors are spinning in pairs ofFigure 1: Rotating pairs of rotors [3]

angular velocity - two rotate clockwise and two rotate counter-clockwise. This is illustrated in Figure 1. Quadcopters have been around for a while and its development can be generally classified into 2 generations. The first generation of designs were done to transport cargo and passengers. However, early prototypes suffered from poor performance [1] and latter prototypes required too much pilot workload, due to poor stability [2].


1


April 7th, 2010

Figure 2: First Generation Design of De Bothezat Quadrotor [3]

The second generation of Quadcopter designs is what the Internet Mobile Robot falls under. This more recent generation consists of designs that are commonly designed to be unmanned aerial vehicles that use electric control systems and sensors to stabilize the aircraft [4]. There are various advantages of the current generation of Quadrotors over comparable scale helicopters. Quadrotors have a simpler mechanism that controls the rotor blades as opposed to a helicopters as mechanical linkages are not required for control of the rotor blades. This simplifies the design of the vehicle, and reduces maintenance time and cost [5]. Also, the use of four rotors allows each individual rotor to have a smaller diameter than the equivalent helicopter rotor resulting in less kinetic energy being stored during flight. This reduces the damage caused should the rotors hit any objects. For small scale vehicles, this makes it safer to interact with in close proximity. Finally, by enclosing the rotors within a frame, the rotors can be protected during collisions, permitting flights indoors and in obstacle-dense environments, with low risk of damaging the vehicle, its operators, or its surroundings [6].


2


April 7th, 2010

1.2

MotivationMultiple implementations of the Quadcopter currently exist. However, most of

them are Radio Control (RC) implementations. Having a Wi-Fi implementation would open up the communications to other possibilities. Adopting this implementation would mean that we would be abolishing the use of the traditional RC remote controller. In lieu of this, a control mechanism needs to be created in the form of an encrypted web browser with a user-friendly Graphical User Interface (GUI). Also, there are various methods to stabilize the RC Quadcopter using accelerometers and gyroscopes. Having chosen to adopt a Wi-Fi implementation, these methods need to be studied and adapted to the Wi-Fi implementation.

1.3

Problem StatementA Quadcopter needs to be constructed and a Wi-Fi implementation needs to be

done to wirelessly manoeuvre and stabilize it via the internet, through a web browser.

1.4

Proposed Solution and AccomplishmentsAs can be seen from Section 1.3, there were four main areas that this project

concerned itself with. The first goal was to get a Quadcopter, comprising of the structure and control logic, constructed. Having done so, a Wi-Fi implementation has to be deployed to control the system. This system then needs to achieve stability and stable manoeuvrability. A user control interface also needs to be developed to control the system in the form of a secure web page.


3


April 7th, 2010

The Quadcopter can be constructed using various component parts from hobby shops. Major parts to look at would be the robots frame as well as the robots flight and control mechanism. A major accomplishment was determining the various required parts and where we could procure them. Having done much preliminary research, a big contribution to the project, the components were purchased from various online stores. The construction of the structure itself was an important accomplishment as it was a necessary component for testing this year. It will be just as important a component for future groups working on this project as it will be foundation on which further applications can be developed. The initial design was constantly modified to meet changing needs. More can be read in Sections 3.0 and 4.0 regarding the selection of components and the design and implementation process. The Wi-Fi implementation can be achieved by using a Wi-Fi module that can communicate remotely with the router. An accomplishment was to find the right Wi-Fi module to communicate with the robot remotely via a router. The WiShield from Async Labs was purchased for this purpose. Commands that the Wi-Fi module receives are communicated to the microcontroller making the robot correspond respectively. This was achieved by learning the various communication protocols (TCP and UDP) and selecting the appropriate protocol. A good understanding of the extensive documentation that came with the module was also necessary and can be seen as an accomplishment since we are now able to write code (client / server) for this module knowing which functions we would utilize. It is an even greater accomplishment that we were able to achieve Wi-Fi controllability as the WiShield is one of the first modules ofDepartment of Systems and Computer Engineering |Carleton University 4


April 7th, 2010

its kind and there was neither support nor a debugging mechanism available. Section 6.0 covers this in detail. Extensive research was done to determine the means to achieve stability of the IMR. Results showed that stability can be achieved using an Inertial Measurement Unit (IMU) and a Proportional-Integral-Derivative (PID) controller. The IMU is the hardware component and the PID controller is the software component that complements the IMU. Two chips work in tandem to provide a Six Degree Of Freedom (6DOF) IMU. The IMU generates raw accelerometer and gyroscope data that are fed to a filter as inputs so that the speed of the motors can be varied accordingly to stabilize the Quadcopter. The Quadcopter is able to stabilize fairly well using this implementation although more testing and debugging needs to be done for this implementation to work flawlessly. Due to time constraints, we were unable to get the Quadcopter completely stable. We were however able to identify why the Quadcopter is unable to completely stabalize itself and this is a key contribution to the project this year so that future teams are saved from looking for causes. Section 5.0 can be referred for further details regarding stability. The PID Controller was just one component of the code required for the proper control of the IMR and hence another major contribution this year was to have a good understanding of the various component functions required for the proper operation of the IMR. A web browser- based GUI was developed using a suite of scripting languages. One accomplishment was to determine which languages needed to be used to build this


5


April 7th, 2010

interface. After much research XAMPP package was selected due to its ability to facilitate various interfacing as well as the potential for further development in the future. After having chosen this package, another major accomplishment was the design of the interface. This design is not perfect but its a good foundation for future designs. It outlines the basic controls currently but this can be improved upon in the future to provide visual feedback, a battery monitor, telemetry etc. according to future development of the IMR. Further details can be read in Section 8.0. Another major accomplishment was the integration of all the code so that the IMR can function seamlessly. The current solution was implemented successfully however it is not the most elegant solution. Section 9.0 has more information with regards to this.

1.5

Overview of the Remainder of the ReportThe remainder of the report consists of documents that constitute research

needed for the IMR project. These documents may be in the form of code, calculations, charts or graphs that illustrate or aid in the understanding of a concept or point. Graphs, figure and tables will be referred to in the pertinent paragraphs, by their corresponding numbers. The List of Figures and List of Tables on pages vi and viii respectively can be referred to as well. Section 3.0 will expound on the process of designing the IMR. It will outline the research and calculations done to come to the final design of the IMR as well as the component parts. Section 4.0 is complementary to Section 3.0 and it will outline theDepartment of Systems and Computer Engineering |Carleton University 6


April 7th, 2010

components chosen and the rationale behind each selection. It will also briefly touch on the construction process of the IMR. Section 5.0 will go into further detail on stability and system controls. The section will delve deeper into topics such as noise-handling, filters and filtering as well as feedback loops and the control mechanism that was used to control the IMR. Section 6.0, Section 7.0 and Section 8.0 constitute materials on the communications component of the project. Section 6.0 looks into how commands are communicated to the IMR. It will look at how the Wi-Fi module has been utilised and how the functions were implemented to communicate wirelessly with the IMR. Section 7.0 describes the client process and Section 8.0 comprises of material on how the user can communicate with the IMR via the GUI created. It also looks at how the GUI was implemented. Section 9.0 explains the software integration process and Section 10.0 provides information on the materials purchased. Section 11.0 concludes the report by looking at what weve achieved this year. It is going to briefly outline the accomplishments and detail various recommendations for future project groups working on the Quadcopter.


7


April 7th, 2010

2.0 The Engineering Project2.1 Health and SafetyDue to the nature of the project, there are many health and safety concerns that one should be aware of before working on such a project. These concerns range from health concerns possibly causing bodily harm to oneself as well as others. Many measures were taken to ensure the health and safety of all members as well as other throughout the span of the project. As with most engineering projects, a lot of time is spent in front of a computer whether it be for research and analysis or for the implementation aspect. Many health concerns can arise if one spends an extended period of time in front of the computer. Sitting with poor posture can cause back pains and wrist pains and staring at a screen for too long may cause ones eyes to strain. As such, it is recommended to take regular breaks to not over strain the eyes and to not cause future back or posture problems. Specific to this project, there are large propellers slicing through the air and a mass flying from point to point. Anytime there are large moving parts that can cause minor to severe bodily harm, one should take safety precautions. Although the

propellers are plastic, the speed at which it is rotating is enough to penetrate the skin. Before commencing with any tests or attaching the propellers and allowing them to spin, one should ensure that there is no one and nothing in the vicinity of the propellers. Also, it is recommended to test in an open space with plenty of room in case one should lose control of the Quadcopter. For testing purposes, the Quadcopter should be

powered using the Power Supply until consistent wireless connection andDepartment of Systems and Computer Engineering |Carleton University 8


April 7th, 2010

communication to the Arduino can be established. This way, there is always a method of turning off the Quadcopter should it start to fly in an unexpected manner. Once the wireless connection can be proven to be reliable, one can use the battery to test the Quadcopter. When working with electrical devices, there can always be a risk of shock or even electrocution. It is extremely important to take precautions when handling electrical components and power supplies. Ensure that everyone that must come in contact with the components know the risks and the proper method of handling the components. Any soldering done should be done with proper equipment and technique. There are guidelines at each soldering station in the labs. One should read and understand how to properly handle the equipment before attempting to solder as this can lead to severe burns.

2.2

Engineering ProfessionalismEngineering is a profession. It is the process of methodically and logically coming

up with a solution for a given problem. As part of the requirements for graduation, our team has formally respected all phases of a development cycle. For it to be considered engineering there must have been some research and analysis of a problem. A solution was then supplied in the form of a Proposal. Further research and analysis was done before the design phase of the project, followed by an interim/progress report (and presentation). Then there was a period of implementation and testing. The team has followed the procedures and are coming to the end of the development cycle for the proposed project.Department of Systems and Computer Engineering |Carleton University 9


April 7th, 2010

Throughout the project, our group made sure our decisions were made responsibly and took precautions to ensure no one in the immediate society would be harmed during the testing and development of the project. Should any doubts arise about the potential safety of anyone, measures were taken to relieve that doubt. Internally, as a team, each group member has handled themselves professionally. In the case of a disagreement, the team would get together and discuss civilly to find a compromise. We have made decisions for the benefit of the progress of the project without attempting anything that may be considered immoral or unethical. Each member took ownership of their duties and ensured that their work was completed as well as it could be.

2.3

Project ManagementOur team started our development by doing some research on our topic of

interest. After some brainstorming sessions for possible ideas, we proceeded with our Quadcopter idea and did some further research and analysis on the topic including possible structure design and components. The brainstorming session really brought together the idea and laid a path for the direction we wanted the project to head in. As the project progressed, different components were selected and designated to a primary that will be responsible for that component from then forward. These topics were chosen to be relatively independent of each other to allow each member to develop an expertise in that topic and allow for a specific area of work while working independently.


10


April 7th, 2010

Team meetings were made regularly to discuss the progress and any problems encountered. Should a problem arise, the team would help in finding a solution or suggest any possible alternatives. Any major changes in the direction of the project would then be discussed and agreed upon. Deadlines for both the department as well as internal team deadlines were emphasized and any upcoming events would be brought up to ensure all team members were aware of their responsibilities for that deadline. Internal team deadlines were made to ensure progress was made in a timely manner. Changes to the project were also made in order to meet deadlines as required. It was decided that only the core components be done for this year, which removed many additional features we previously wanted to implement. We also created an online group with Google Groups to allow every member to add and access any documents another member may be using. This helped with organizing documents and reports as well as allowed us to share any useful information with the group in a well managed, designated area. It quickly became a good source for referencing any previous discussions and documents.

2.42.4.1

Individual ContributionsProject Contributions The area of expertise and designated work was assigned as stated below. The project

could be divided into four high level components: 1. Structural Analysis and Design (Rahul Bura) 2. Stability and Manoeuvre (Hue Vo) 3. Wireless Communication (Mohamed Chande Server, Vinayak Goge Client) 4. User Interface and Webpage Design (Vinayak Goge) Department of Systems and Computer Engineering |Carleton University 11


April 7th, 2010

The work was assigned as such in order to allow for any additional required independent work to be done outside of team meeting times. Much of the work in the development cycle, from Research and Analysis to Implementation to Component Testing, were done as a team where suggestions and any further knowledge of the topic by other teammates could help progress the project more efficiently. Much of the Research and Analysis was done by all members of the team with Rahul Bura leading the structural and component selection. During integration of the project, all members were equally involved, regardless whose component was being integrated at the time. For formal presentations, all members contributed in different components throughout the presentation preparation period. 2.4.2 Report Contributions As with the Project Contributions, as previously stated, this report was written by the person with the most expertise of the topic throughout the project, as much as possible. The formal contributions are as follows: Section 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 Author Rahul Bura Hue Vo Rahul Bura Rahul Bura Hue Vo Mohamed Chande Vinayak Goge Vinayak Goge Mohamed Chande, Hue Vo Rahul Bura, Vinayak Goge Rahul Bura, Mohamed Chande, Vinayak Goge, Hue Vo Editing and Proof Reading Mohamed Chande Vinayak Goge Hue Vo Mohamed Chande Rahul Bura Vinayak Goge Mohamed Chande Hue Vo Rahul Bura Mohamed Chande Rahul Bura, Mohamed Chande, Vinayak Goge, Hue Vo

Table 1: Report Contributions


12


April 7th, 2010

3.0 Robot DesignRobot design is one of the key components of the project. The IMR consists of the structure and logic elements. The structure designs were considered based on manoeuvrability, aerodynamics, and cost of component parts and feasibility of the design. The logic elements were chosen on the design decisions made to address the problem statement. The design process took into consideration key components that include structure of the robot, communications, the stability mechanism, the controls mechanisms and power.

3.1

StructureThe structure is one half of the IMR. The structure houses the microcontroller

and also has the flight mechanism integrated. Designing the structure design required various design considerations including the number of motors and frame design.3.1.1 Number of Motors

Technically, any number of motors can be mounted to achieve flight. However, with every new motor mounted, the weight of the robot increases. Weight is obviously a concern and we want the IMR to be as light as possible. There are two types of weightweight of the IMR with the basic components mounted components and weight of extra mounted parts, also known as payload. It is important to note the correlation between the weight of the IMR , its ability to carry that weight and the number of motors. Table 2 tabulates these factors to help with the design decision. 2 designs were considered - the Quadcopter (4 motors) and the Octcopter (8 motors) as is reflected in the table.


13


April 7th, 2010

The metric used to determine the design is flight time and a 2100mAh battery was used as a constant power source to determine the flight times for all the payloads of each of the two designs. It is assumed that each design is able to carry its own structural weight and hence they are ignored so only the payload is tabulated.

4 Motors Design Pay Load(oz) (Quadcopter)

8 Motors Design (Octocopter)

0 8 16 24 32 40

19.9 12.6 9.2 7.3 -

11.7 Flight Time (Mins) 9.2 7.5 6.3 5.3 4.6

Table 2: Flight Times for 2 Different Designs

As can be observed, with no payload, the Quadcopter can achieve flight for about 20 minutes and the Octocopter can achieve flight for about 12 minutes. This demonstrates the effect of the weight of Octocopters frame. It has a decreased time of flight compared to the Quadcopter without a payload. The maximum payload that the Quadcopter can carry is 24oz for a period of 7.3 minutes. This is in contrast to the 40oz that the Octocopter can carry due to the increased number of motors. However, it can only sustain flight for about 4.6 minutes which is not ideal. We also expect that the maximum payload we will have would be less then 16oz. Hence looking at the payload of 16oz, the Quadcopter offers a better performance and we settled on the four-motor design for our IMR.Department of Systems and Computer Engineering |Carleton University 14


April 7th, 2010

3.1.2

Frame

The frame of the Quadcopter was next in the design of the structure. Numerous materials were researched with the criteria that the material be durable and light. Some of the alternatives considered were aluminum, balsa wood and carbon fibre reinforced plastic. Aluminum is a soft, durable, lightweight and malleable metal that is easy to work with. It has about one-third the density and stiffness of steel making it significantly lighter [7]. Balsa wood is one of the lightest varieties of wood available and strong for its weight, pound for pound. Its fairly malleable without compromising its strength [8]. Carbon fibre reinforced plastics are composite plastics that have been reinforced with carbon fibre to provide a high strength-to-weight ratio. The density of carbon fiber is also considerably lower than the density of steel, making it ideal for applications requiring low weight [9]. The design of the frame itself was based on other models. It was decided that well have a cross-shaped frame with a motor mounted on each of the four arms. The intersection in the middle would have the logic and power units mounted on it. The illustration in Figure 6 shows the preliminary proposed structure.

3.2

CommunicationsHow we communicated with the robot was a key consideration. Various

alternatives we looked at including ZigBee, Wi-Fi and Bluetooth. After careful analysis of the specifications and capabilities of each technology, we decided on implementing WiFi for our robot design.


15


April 7th, 2010

Bluetooth technology however was ruled out earlier in the course of development of the project mainly because of the cost to transmission range trade-offs. Bluetooth modules can provide transmission ranges of 1m, 10m and 100m. For our project purposes, the ideal minimum range was considered to be 100m. With this range, Bluetooth technology is very expensive. Furthermore, compared to Wi-Fi and ZigBee, Bluetooth networking is slower and this feature is undesirable in real-time applications and applications that are sensitive to network delays. ZigBee, Wi-Fi standards and their corresponding Radio Frequency (RF) modules provide specifications that meet the requirements of the project. As to how Wi-Fi was selected over ZigBee, a detailed explanation is provided in Section 6.2.

3.3

Flight and StabilityFor flight, four pairs of propellers and motors are required. There are numerous

motors available that are powered by DC or AC sources. We were particularly interested in DC motors since we wanted to power components using battery cells. DC motors run on DC electric power and there are various types that include the Brushed DC motors and Brushless DC Motors. Brushed motors refer to the classic DC motor that has a wound rotor with a split ring commutator which periodically reverses the current direction between the rotor and the external circuit, and a magnet stator. [10] An electrical power source is connected to the rotor through the commutator and its brushes providing current flowDepartment of Systems and Computer Engineering |Carleton UniversityFigure 3: Wound Rotor [10]

16


April 7th, 2010

and subsequent electromagnetism.

With the commutator switching currents

periodically as the rotor turns, the magnetic poles of the rotor are prevented from ever being fully aligned with the magnetic poles of the stator field causing the rotor to spin indefinitely. Essentially, stationary metal contacts that brush against moving metallic contacts are used to transfer electrical energy to coils on the rotor. However, the brushed implementation has a number of limitations. Main ones include a limit to the maximum speed of the machine and the need for replacing the brushes. The alternative is the brushless motor. The brushless motors implementation differs from that of the brushed motors. A brushless motor consists of stationary coils and a rotating magnet. The need for brushes to provide currentFigure 4: Brushless Motor Design [11]

to the moving rotor is eliminated and instead an

external electronic controller is used to power up the stationary coils, which are grouped in phases, causing the magnet to rotate. [11] Section 4.3.2 contains further discussion on the electronic controller. Figure 4 shows the brushless motor design. The star-shaped component with the blue coils is the rotor and the disk in the top right corner has the permanent magnet in the shape of a ring and it rotates around the rotor, about the centre. This design differentiation provides several advantages over the brushed motors.


17


April 7th, 2010

Research showed that an ultimate combination of motor and propeller is needed to be used for optimal thrust power. The appropriate combination was chosen as can be seen in Section 4.2. A 6 degree of Freedom (DOF) Inertial Measurement Unit (IMU) was required for stability and manoeuvring the Quadcopter. An IMU is an electronic device that measure and reports velocity, orientation and gravitational forces using accelerometers and gyroscopes in tandem. Accelerometers measure acceleration relative to freefall and gyroscopes measure or maintain orientation, based on the principles of conservation of angular momentum [12]. Since the Quadcopter is going to be flying in a threedimensional space, an IMU that considered the 6 degrees of freedom was essential. The 6 degrees of freedom refers to the fact that the rigid body is able to move about the X, Y and Z axes independent of each of the 3 axes and of the rotation about any of the 3 axes. This would enable the Quadcopter to move forward, backward, up, down as well as left and right in three-dimensional space.

3.4

Flight ControlTo control the Quadcopter, two alternatives were considered. We could have

used a Field-Programmable Gate Array (FPGA) board or a microcontroller. A FPGA is an integrated circuit that contains programmable logic components known as Logic Blocks and a hierarchy or reconfigurable interconnects that allows the blocks to be connected. These configurations can be programmed using Hardware Description Language (HDL). However, an FPGA would have been harder for us to work with as we were not as familiar with it as we are with our other alternative the microcontroller.Department of Systems and Computer Engineering |Carleton University 18


April 7th, 2010

Microcontrollers are computer systems on a chip. Microcontrollers have a microprocessor and peripheral functions like a relatively simple clock, timers, I/O ports and memory implemented on one chip. Figure 5 illustrates this. Microcontrollers are designed for smaller or more dedicated applications and hence they may have lower clock rate frequencies enabling lower power consumptions. This would be perfect for battery-powered applications such as our Quadcopter. Microcontrollers were originally programmed in assembly language (low level language communication at machine level) but various high level programming languages are now in use as well. Such languages can be specially written to be microcontroller-specific or be versions of general purpose languages such as the C programming language. Compilers and environments may be tools that are provided by the microcontroller vendors to program or debug the microcontroller.

Figure 5: Typical Microcontroller

Various microcontroller options were considered as well as there are numerous manufacturers. Ultimately the cost, form factor and ease of programming are the criteria that would determine the choice.Department of Systems and Computer Engineering |Carleton University 19


April 7th, 2010

3.5

PowerThe Quadcopter needed a sustainable and portable power source to power the

control unit and the motors. Different types of rechargeable batteries were researched and a number of chemical compositions were taken into consideration. Nickel Cadmium (NiCd), Nickel Metal Hydride (NiMH), and Lithium Polymer (LiPo) cells are currently the most commonly used, but each needs to be charged, discharged, and stored differently. On top of that, each model may require a different cell count or battery configuration as well. Nickel Cadmium or NiCd batteries are less common now but they are cheap. These batteries have cons as well however. NiCd batteries need to be fully discharged after each use as failure to do so would mean that for future discharge cycles, they will not discharge to their full potential. NiCd batteries also have a low energy density the capacity per weight. Nickel Metal Hydride (NiMH) batteries have numerous advantages over the NiCd batteries. NiMH cells offer higher energy density and dont have the same performance issues attributed to improper discharge practices as NiCd batteries do. The latest cells are the Lithium Polymer (LiPo) cells. LiPo cells offer higher better discharge performance as they provide better consistency compared to NiCd and NiMH cells. LiPo cells also offer a significantly higher capacity for their weight; a cell may have twice the capacity for half the weight of a similarly performing NiMH cell. Hence, LiPo cells can achieve higher voltage and energy density. LiPo cells need to be monitored when being charged however. This is the major deterrent when it comes to adopting


20


April 7th, 2010

this technology. Overcharging can cause the cells to be potential major fire hazards given the amount of energy packed into such a small space. The Lithium Polymer battery was chosen in the end due to the advantages mentioned above.


21


April 7th, 2010

4.0 Hardware Components and ConstructionHardware components were chosen from the alternatives presented in Section 3.0. It was important to bear in mind the weight of each component. There were other decisions that needed to be made in the process of choosing an individual component. These decisions were made to ensure compatibility with other component parts. This will be discussed in each of the sections.

4.1

Frame and StructureCarbon fibre reinforced plastic (Carbon Fibre for short) rods were used for the

arms of the robot. Carbon Fibre Rods are light and proved to have a higher strength-toweight ratio when compared pound- for-pound with the other alternatives. Carbon fibre also has high tensile strength, low thermal expansion. However, it is relatively expensive when compared to the other alternatives but the small price discrepancy was almost negligible looking at the advantages Carbon Fibre provided, especially since it is significantly lighter and durable compared to the alternatives mentioned in Section 3.1.2. The sketch in Figure 6 shows the proposed basic frame of the IMR. Aluminum was chosen for the square base initially as can be seen in Figure 7. The figure shows the initial structure with the Carbon Fibre rods and aluminum square base. The total weight of the fibre rods, the aluminum base and the miscellaneous screws and nuts used to secure the components is 3.2oz.


22


April 7th, 2010

Figure 6: Sketch of Proposed Structure

Figure 7: Initial Frame


23


April 7th, 2010

4.2

MicrocontrollerThe Arduino Mega was chosen because of its greater memory, processing power and number of ports. It has 54 digital input/output pins, 14 of which offer Pulse Width Modulation (PWM) that is required to control the motors, 16 analog inputs that provide a 10bit resolution each, and 4

Serial UARTs. The Arduino Mega is a microcontroller board based on the ATmega1280 microprocessor. It has an operating voltage of 5V, input voltage range from 7V to 12V, 128KB of Flash Memory for storing code, 8KB of SRAM, 4KB of EEPROM and a clock speed of 16MHz. It was a cheaper alternative to the other options considered. The microcontroller is widely adopted and hence there is more support for it. There are numerous shields that can be mounted on to it for added functionality. Examples include the Aeroquad Shield V1.7 and the WiShield 2.0 that will be discussed in later sections. The Arduino Mega can be programmed with the Arduino Software provided free by the developers. The Arduino Integrated Development Environment (IDE) is written in Java and made for the Processing programming language. It includes a code editor with features such as syntax highlighting, brace matching, and automatic indentation, and is also capable of compiling and uploading programs to the board with a single click.The IDE also comes with a C/C++ library that can be used to simplify I/O operations. Arduino programs are written in a language akin to C/C++ and hence it is something that we are familiar with [13]. The Arduino Mega contributes a weight of 1.5oz to the IMR.Department of Systems and Computer Engineering |Carleton University 24


April 7th, 2010

4.34.3.1

Flight and StabilityPropeller and Motor Combination Configuration

Brushless DC motors were ultimately chosen to provide thrust power to the IMR. Brushless DC motors have a higher efficiency of 85% to 90% as opposed to the 75% to 80% efficiency of brushed motors. This results in reduced noise, longer lifetime and more power. These motors provide superior power-to-weight ratios and are very light. To pick the right motor model, the propeller was chosen first. The light EPP1045 propeller was chosen as they are widely adopted in other implementations of the Quadcopter. With this design decision, the matching motor for optimal thrust was sought and research yielded Table 3 below.

Table 3: Matching Motors for EPP1045 Propeller

The table shows the comparison of the performance of the propeller with various motors. The criteria in consideration are cost, weight of the motor, maximum thrust achieved in pounds and the amount of power drawn per pound at maximum


25


April 7th, 2010

thrust. After the analysis was done, what it came down to was availability. The only motor available was the Towerpro 2410-09 Open Base Brushless Motor. The statistics show that it is a cost effective option and that it is really light at 2.05oz. It could provide better thrust but it is one of the more power-efficient motors. With a total of four such motors on the IMR, they will contribute a total weight of 8.2oz. The Amp rating is 13.5A. The positioning of the motor also contributed to the stability. The closer the motors are to the centre, the more stable the IMR is. However, this compromises the manoeuvrability of the IMR as the Quadcopter responds better to change in directions when the motors are further apart. This increases the signal noise fed into the IMU due to increased vibration. Extensive testing was done to determine the perfect distance of the motors from the centre for the right balance between noise and manoeuvrability. The propeller/motor configurations also differed. As mentioned in Section 1.1, there are 2 motors that spin clock-wise and two that spin counter clockwise. The right propeller had to be mounted on the corresponding motor to achieve this. The motors were wired to Electronic Speed Controllers (ESCs) in 2 configurations to ensure the clock-wise and counter clock-wise rotations of the motors. ESCs are discussed in Section 4.3.2.4.3.2 Electronic Speed Controller (ESC)

Since a Brushless Motor was chosen, an Electric Speed Controller was required. A brushless motor controller or brushless ESC (Electronic Speed Control) is used to vary the speed of a brushless motor. These function as an interface between the motor and


26


April 7th, 2010

the battery. Controlled by the microcontroller, the brushless ESC provides variable power to the motor allowing proportional speed adjustments. The microcontroller sends PWM signals with different duty cycles to vary the speed of the motor rotation. Unlike a brushed motor, power cannot be directly applied to a brushless motor. Instead, the speed control intelligently powers each phase of a brushless motor in sequence, causing it to rotate. The Tower Pro w18A Mag8 Digital Brushless Motor ESC was chosen. This component has a weight of 0.705oz, an Amp rating of 18A and operates ideally at a voltage of 6V 12V. These are considerations that are important to note. The voltage information needs to be noted for the purpose of selecting the battery. The the total weight contributed by the 4 ESCs is 2.8oz. The Amp Rating of 18A indicates the maximum current that the ESC is able to provide continuously to a motor and it is better to have an ESC with a higher continuous current rating to ensure that the ESC is able to handle the power requirements of the motor. The Amp rating 18A is good given the 13.5A rating of Tower Pro Brushless Motor 2410-09 chosen.4.3.3 Six Degrees of Freedom (DOF)

After much research, two chips from SparkFun were chosen to provide the 6 DOF for the Inertial Measurement Unit. The first is a 5DOF IMU combo board that incorporates the IDG500 dual-axis gyroscope and the ADXL335 accelerometer on one single chip. This board enables the 5 axis of sensing (Roll, Pitch, X,Y,Z) in less than 1 square inch and weighs less than 0.07oz. The second board is the dual axis IXZ-500 gyro. This senses the angular velocity on the X and Z axes. This board thus provides YawDepartment of Systems and Computer Engineering |Carleton University 27


April 7th, 2010

information and it complements the first board that provides Roll and Pitch. These two chips can be mounted onto the Aeroquad Shield. The Aeroquad Shield is a printed circuit boards (PCB) that was made for the Arduino. It was created to mount the two chips as well as to interface with the ESC connections. The total weight contributed by this set-up is 0.75oz.Figure 8 : Aeroquad Shield

4.4

CommunicationsThe WiShield 1.0 was the chosen Wi-fi module from Async Labs. This is perfect

for the Arduino-based projects and is a shield module that can be directly mounted on the Arduino Mega. This shield provides 802.11b connectivity and is a direct drop-on plug-and-play solution. It has a 16Mbit serial flash onboard to store web pages and other data. This space can also be used for storing sensor type data that can be downloaded in the future. It supports both infrastructure and ad hoc wireless networks which can be useful in the testing phase. Further research can be found in Section 6.0.

4.5

BatteryThe Lithium Polymer (LiPo) cells were chosen since they clearly provided better

performance compared to the NiCd and NiMH battery types. It is really light and has a good energy density. This is perfect for the project as more power would be provided from a battery pack that doesnt weigh too much. A 11.1V battery was chosen and this is a good volatage rating recalling that the ESCs has a input voltage range of 6-12V and the microcontroller has an input voltage of 7V to 12V.Department of Systems and Computer Engineering |Carleton University 28


April 7th, 2010

4.5.1

Flight Time and Battery Power Dependancy

Given that the Quadcopter would have its own weight as well as other payloads, we needed to do an analysis to see how much load a certain battery can provide and how much time the battery will provide for flight given the pay load. The various components and the corresponding weight calculated in previous sections have been tabulated in Table 4. The table indicates the weight of the Quadcopter with its basic component parts. This weight will be considered the IMR weight. Anything on top of this weight will be considered payload. Measurements were taken using a 2100mAh battery again to see how much payload this configuration can take and to see if it is consistent with the Table 2 in Section 3.1.1. The graph generated can be seen in Figure 9.

Component Base/Frame Microcontroller EE1045 Propellers and Prop Savers 2410-09 Motors Motor Mounts ESCs Bindings and Miscellaneous Parts Battery Total Weight

Weight(oz) 4.2 1.5 1.1 8.2 0.6 2.8 0.4 5.2 24.0

Table 4: Total Weight Calculation


29


April 7th, 2010

Flying weight vs Battery Life25 20 Battery Life ( Mins) 15 10 5 0 0 10 20 30 40 50 60 70 Flying Weight in Ounces (oz) 12.6 10.5 2100mAh 7.5 7.3 6.4 5.5 4.6 19.9 17.1

9.2

Figure 9: Graph of Flight Time versus Battery Life

The results above show that with no payload (i.e. 24oz), 19.9 minutes of flight time can be achieved. The graph was extrapolated to predict the flight time for heavier payloads but it might not necessarily be realised in reality.

4.6

Design Implementation and Final StructureHaving considered the numerous factors to choose component parts, the

eventual implementation of the design was both mechanical and electrical in nature. Basic mechanical work was done with regards to the construction of the structure. Minor cutting of the Carbon Fibre rods and drilling was done. Two acrylic pieces were added as a final design decision to provide two different levels. The bottom level is used to store wiring and the battery and the upper level is where the logic unit is situated. Minor electrical skills were also required to solder the component parts and wire them. Figure 10 shows the top level wiring diagram andDepartment of Systems and Computer Engineering |Carleton University 30


April 7th, 2010

Figure 11 shows the final design.


31


April 7th, 2010

Figure 10: Wiring Diagram for IMR


32


April 7th, 2010

Figure 11: Final Design


33


April 7th, 2010

5.0 Stability and ManoeuvreThere are obviously many factors that affect the flight of the Quadcopter. The collaboration of components as previously described is a core factor that determines whether the Quadcopter can take flight or not; the sections to follow describe ways to stabilize the Quadcopter when it is in the air as well as the constraints on the system that will allow us to steer the Quadcopter.

5.1

Filtering NoiseA clean analog signal is almost impossible to get. It is always littered with noise

whether it is from the AC signal from the outlets that add 60Hz noise or mechanical motors that cause vibrations. This noise usually cannot be completely removed but needs to be minimized in order to get the data that we need. A Second Order

Complementary Filter was used to remove the noise from the accelerometers and gyroscopes caused by a variety of sources from electrical to mechanical.5.1.1 Noise Reduction

To obtain the desired signal, it is theoretically possible to use a filter and isolate the desired signal. Reality is, it is harder than it sounds. There are many uncontrollable variables ranging from predictable (such as 60 Hz AC noise) to unpredictable (such as mechanical vibrations and wind) which may add noise to the system and make it more difficult to isolate for the desired signal. It is extremely difficult to eliminate all noise from any given signal. The noise is mixed in with the desired signal and simply applying a filter may not be able to remove the noise without also potentially removing the signal. In fact, it is impossible to remove just the noise without attenuating the signal ifDepartment of Systems and Computer Engineering |Carleton University 34


April 7th, 2010

the frequencies overlap or if there are common frequencies. For this reason, the Signal to Noise Ratio (SNR), the power ratio between the signal and noise, has more value in noise reducing processes and is more frequently used in practice. For the purposes of this project, moderately controllable variables such as the weather and wind are removed by testing in an open space that is indoors with relatively low draft. Mechanical and electrical noise from the equipment is monitored and removed as much as possible without distorting the desired signal itself using a Second Order Complementary Filter.

5.1.2

Second Order Complementary Filter

A complementary filter is a filter with a derivative feedback through the filter. [14] The order of a filter determines how many components that is required. The high level design for a second order filter requires two integrals from the inputs to the output. Therefore, a Second Order Complementary Filter is one with two integrals, with the value of the output being fed back between the first and second integral. A block diagram of the Second Order Complementary filter can be seen in Figure 12.

Figure 12: Second Order Complementary Filter [15]


35


April 7th, 2010

Theoretically and through experiments, it can be seen that the Second Order Complementary Filter performs drastically better than a First Order Complementary Filter and without wasting too much more computation time. It can be seen that a Second Order Complementary Filter removes the noise from the sensors quite well and leaves a relatively clean signal as compared to the signal received (refer to Figure 13).

Figure 13: Second Order Complementary Filter [15]

5.2

Feedback ControlAs with most control systems, the time it takes for the system to reach Steady

State is of utmost importance and is part of the tradeoffs that must be taken into consideration. A system must respond in a timely manner. There are many factors that


36


April 7th, 2010

may affect the response time but a couple of major factors affecting Transient Response: Computing/executing speed of system, Communication delay. Computing speed of system is dependent on the speed of the microcontroller used, but as processing speed is getting faster and faster, finding a microcontroller that is fast enough has become less of a problem. Execution at the optimal speed and with the optimal response is still very important. There are methods used in control

engineering that optimize the speed at which the system reaches steady state. Communication delay is inevitable with any system. We can only ensure that we allow enough time for the message to be received before re-sending a signal. considerations will be discussed in Section 6.0 Wireless Communication.5.2.1 Six Degrees of Freedom (DOF)

Further

The principle is that a rigid body in three dimensional (3D) space has six independent ways it can move (or six DOF). The movement can either be translational or rotational about the X, Y, and Z axis or combinations thereof (Figure 14). In flight terms, the axes are called Roll, Pitch, and Yaw. The Roll is tilt from side to side. The pitch is the elevation of the front. The yaw is the direction of movement relative to the desired projection, also referred to as side slip. Constraints are needed for each DOF in order to restrict or enable movement in 3D space.


37


April 7th, 2010

Figure 14: Possible independent movements in 3D space [16]

5.2.2

System Control Theory

To control an object in 3D space that is free to move about, all six DOF must be monitored for unplanned directional changes and adjusted accordingly. Adjustments can either be done manually or with an automated control system to ensure the object is of the correct orientation and angle. Control Systems can be classified in to three categories: Open Loop, Feed-Forward, and Feedback. [17] Open Loop Controllers, also referred to as Non-Feedback Controllers, are useful for systems where the system inputs are directly related to the system outputs and the desired system state, usually by a mathematical formula. Feed-forward Control Systems are generally used when the effect of an input or command to the system produces a predictable output. Feedback Systems (or Closed-Loop Systems) are causal systems that are mostly used in systems that require adjustments and possibly machine learning. The output of causal systems depends on the previous output(s) of the system. For

Feedback System, the output is determined by comparing the previous output signal and the reference signal. [17]Department of Systems and Computer Engineering |Carleton University 38


April 7th, 2010

Figure 15: Types of Control Systems: (a) Open Loop (b) Feed-Forward (c) Closed Loop [18]

As seen in Figure 15, different types of systems will react differently to disturbances to the system. From the description of the project and from the

descriptions of the types of systems, a Closed Loop Feedback System is the most suitable for a Quadcopter that must adjust according to the sensors that determine its orientation and direction. There are two types of Feedback Systems: Positive Feedback Systems and Negative Feedback Systems. [17] Positive Feedback Systems are causal systems that amplify any small disturbances to the system. A small perturbation can cause the system to grow

relatively quickly. This is useful when the output of the system needs to be magnified or


39


April 7th, 2010

enhanced. [17] In the case of a Quadcopter, a Positive Feedback Systems would ultimately magnify any noise present in the system and destabilize the system. Negative Feedback Systems are also causal systems, but instead of amplifying any small perturbations they will make up for any discrepancy between the desired output (according to the input) and the actual output so that the system can reach the desired output. [17] The output of the system is the signal that is fed back into the system. The difference between the desired signal and the signal fed back is the command error and will determine how the system should react. If the desired signal and the feedback signal match, the command error is zero and the system has reached the desired state. Along with how a system will reach a stable state is how quickly a system can reach that steady state. The time it takes to reach steady state is dependent on the damping of the system. A system can be described to be underdamped, critically damped, or overdamped. An underdamped system will overshoot the steady state value and oscillate about the steady state value before it reaches steady state. The more underdamped the system, the higher the amplitude and number of the oscillations. A critically damped system is a system that reaches steady state as quickly as possible without overshooting the steady state value. An overdamped system will reach steady state without overshooting the desired value but over a longer span of time. The dampness of a system is represented by the damping constant, . If 0 < < 1, the system is considered to be an underdamped system which means the transient


40


April 7th, 2010

response is oscillatory. If = 1, the system is described as critically damped and if > 1 the system is overdamped, as seen in Figure 16. [17]

Figure 16: How the damping constant affects the time it takes to reach steady state [19]

Because of the nature of the system for a Quadcopter, manual adjustments are very difficult as the response must be instantaneous, yet human reaction is only so quick. Thus, to steer the Quadcopter, we must resort to a mechanical-electrical

solution: Accelerometers and Gyroscopes. These mechanisms will help direct and control the direction of motion of the Quadcopter by outputting any possibly unwanted changes to the system.5.2.3 Proportional, Integral, Derivative Controller

A Proportional, Integral, Derivative (PID) Controller contain three terms (controllers) that are used to control the gain and thus the overall response time of the system: Proportional, Integral and Derivative. PIDs calculate the difference between the


41


April 7th, 2010

output value and a set desired input value and use it to calculate and control the gain of the system and maximize the response time. The Proportional Controller determines the reaction to the current error, the Integral Controller sums the recent errors and provides a better steady state response, the Derivative Controller enhances transient response but uses the rate at which the error has been changing and therefore can be unpredictable and cause a lot of jitters. Together, the three terms are used to control the gain and response time of the system. The tradeoff for speed is how much the system will oscillate. For a system like the Quadcopter, too much oscillation is

undesirable so the gains will be set accordingly. The time domain transfer function can be seen and derived fromFigure 17. The S-Domain transfer function of a PID is: H(s) = KP + KI/s + KDs = KDs2 + KPs+ KI. Using the Sdomain transfer function, one can determine where the poles and zeros of the system are. The poles are at the denominator factors and the zeros are at the numerator factors for a fully factored transfer function. For a system to be stable, all poles must be in the open left half plane (OLHP). If there are any poles in the open right half plane (OLHP), the system is considered unstable. Any poles on the imaginary axis is

considered marginally stable (or pure oscillatory). The closer the poles are to the imaginary axis, the quicker the response, but at the same time the system can become more oscillatory which is undesirable for a Quadcopter. Over time, it is expected that the error in the feedback loop approach zero: the steady state of the system with the current input. [17]


42


April 7th, 2010

Figure 17: PID Controller loop [20]

5.2.4

Feedback Control Mechanism: Inertial Measurement Units

Inertial Measurement Units (IMU) are widely used to manoeuvre and control the direction of moving objects from airplanes to satellites to rockets. It consists of a combination of accelerometers and gyroscopes. As previously discussed, for a system such as a Quadcopter, the system would need to constrain all six DOF in order to fully navigate the Quadcopter predictably. The three accelerometers can be used to

constrain the translational motion along the three independent axes and the gyroscopes can be used to constrain the rotational motion about each of the three independent axes. The accelerometer measures the acceleration relative to the frame and acts as a motion sensor to determine the direction and orientation of movement. It would firstly need to be zeroed for a certain plane. That plane would then be the position of rest. The output of an accelerometer is simply a voltage level that is increased if the accelerometer is tilted to one direction, or decreased if it is tilted in the other direction. From the output of the accelerometer, one can determine the angle at which the Quadcopter is currently at. With three accelerometers placed perpendicular to each other, one can determine the angle for each of the axes.Department of Systems and Computer Engineering |Carleton University 43


April 7th, 2010

The gyroscope maintains the direction by detecting the change in orientation. Gyroscopes maintain the orientation using the concepts of conservation of momentum. The gyroscope continuously spins about an axis. Once the spin is axis is skewed, the output will tell us how the gyroscope is moved and what must be done in order to balance it out again. As with the accelerometer, the gyroscope also has to the zeroed to determine the rate of change at the desired constant value (usually at rest). The gyroscope would then output a certain voltage level when a change in orientation is detected and this value can then be used to adjust the orientation accordingly. Using the one device without the other is simply not sufficient. Both devices must be used in tandem in order to control both the orientation and direction of movement of the Quadcopter. PIDs will determine how fast it will reach that steady

state while accelerometers and gyroscopes determine the direction and magnitude of adjustment required to reach steady state due to the inputs. Both the gyroscopes and accelerometers can be found in compact forms in an integrated chip the size of a quarter.5.2.5 Feedback Control Loop

Now that we have generally described all the desired components that will help stabilize and manoeuvre the Quadcopter, we can put the components together. A generic form for a feedback control loop with the controller and sensors is:

Figure 18: Feedback loop including system, controller and sensor configuration [21]


44


April 7th, 2010

We can replace the generic controller with our controller of choice: the PID controller. Similarly, the System is our Quadcopter (consisting of the structure, motors, propellers, etc.) and our output sensors for the feedback loop are the three accelerometers and three gyroscopes. Our feedback control loop design for the

Quadcopter can be represented by the block diagram below.

Figure 19: Feedback Control Loop with control components

From Figure 19, one can follow the feedback loop to determine the sequential output of each component. The difference between the sensor outputs and the

reference signal is determined which is then used by the PID to determine the optimal input to the Quadcoptor so that the output of the system is as desired (the system output being the flight orientation and direction). The accelerometer and gyroscope outputs will then be fed through the filter and the measured command error will be again calculated. The filter was added after the sensor data and before calculation of the measured command error so that it can remove the noise from the sensors before using the signal in any component of the system. It should be understood that each control component uses a different controller. The Quadcopter system as a whole can be split into different subsystems. The PID values for the Roll control may be different than that of the Pitch, Yaw or Throttle. In addition to those PID controllers, there are also different PID controllers for the autoDepartment of Systems and Computer Engineering |Carleton University 45


April 7th, 2010

levelling (or stabilization) of the Quadcopter to try to optimize the response to a disturbance to the system.5.2.6 Flight Tuning Using Ziegler-Nichols Rules

Many systems can be easily modelled mathematically and an analytical approach to determining the values for the PID controller can be used. For systems that are difficult to model mathematically other approaches such as the Ziegler-Nichols Rule can be used to obtain an educated estimation for the PID which can later be fine tuned further. [17] Type of Controller KP P PI PID 0.5KCR KI 0 KD 0

0.45KCR 1.2/PCR 0 0.6 KCR 2/PCR 0.125PCR

Table 5: PID gain using Ziegler-Nichols Tuning Rule [17]

To use the Ziegler-Nichols Rule, one must find the critical gain, KCR, and critical period, PCR, of the system. To find those two values, one must set K D = 0 and KI = 0 for the PID Controller transfer function: H(s) = KP + KI/s + KDs. Adjusting only KP, one must find the value that causes the system to exhibit sustained oscillations. The gain at which the system exhibits sustained oscillations, is known as the critical gain. Using the frequency at which the system oscillates, one can find the critical period. Then

according to Table 1, one can choose the desired values for the PID controller. According to the Ziegler-Nichols Rules, this PID should exhibit close to optimal response for the system, with perhaps minor fine tuning around these gain values.


46


April 7th, 2010

5.35.3.1

Flight Configuration and SimulationFlight Configuration Methods and Tools

There are many variables that must be configured in order to control the Quadcopter in flight. These variables are stored in EEPROM and read before and throughout operation. There are two methods of setting these variables: Using the Serial Monitor or with the aid of AeroQuad Configurator v1.2. The Serial Monitor is built into the Arduino development environment. It allows for serial communication between the Arduino board and the user computer via USB. This communication medium is used to upload programs as well as to debug via print statements and send and request flight configuration parameters.

Figure 20: Motor Command outputs (S) during simulation of flight with Serial Monitor

The commands to send and request flight parameters can be found in SerialTelemetry.pde and SerialCommand.pde, respectively. The commands need to be


47


April 7th, 2010

entered in the command line of the serial monitor. To send or request a command, enter the letter corresponding to the command, select send and the values will appear in the Serial Monitor output screen below it. Refer to Figure 20 for the outputs using the command S which requests the motor commands and various other flight parameters separated by a comma. In the same manner, one can simulate flight (without propellers attached for safety reasons) by tilting the Quadcopter and sending the Q command to receive continuous sensor data to monitor the change as the Quadcopter is being moved Figure 21.

Figure 21: Sensor Data output (Q) during simulation of flight with Serial Monitor

Alternatively, AeroQuad Configurator is an open source software developed by AeroQuad to test and adjust flight parameters of the Aeroquad before flight. This software can also be used to set various parameters that control the flight of the Quadcopter such as the PIDs, filter time constant, transmitter/receiver sensitivity,Department of Systems and Computer Engineering |Carleton University 48


April 7th, 2010

levelling limits, etc. Its Graphical User Interface (GUI) is very straight forward and easy to understand and use (refer to Figure 22). The values can be entered into the corresponding box and updated by selecting the Update button at the bottom right corner.

Figure 22: AeroQuad Configurator GUI with updatable flight parameters

In a similar manner, values such as the PID outputs, motor speeds, sensor values, etc. can be plotted and seen in graphical form with the AeroQuad Configurator Figure 23. The continuous time line graph helps one to visually see the oscillations occurring and the sizes of the oscillations.


49


April 7th, 2010

Figure 23: Various sensor outputs with AeroQuad Configurator GUI

There is known to be a few bugs with this tool when it comes to setting these parameters. Even if it says the parameters have been updated, it may not necessarily be so. The values should always be double checked with the Serial Monitor before flight to ensure the values were indeed written to EEPROM.5.3.2 Pre-Flight Tests

Before the Quadcopter takes flight, there is a list of tests that should be executed. This is for the safety of everyone involved and everyone in the vicinity of the Quadcopter while it is flying. It is to ensure that the Quadcopter does not fall out of the sky and or crash into anyone. This could prevent damaging the Quadcopter as well as prevent causing any bodily harm to oneself and others. 1. Increase the throttle to a point where all motors are spinning.


50


April 7th, 2010

2. Tilt the Quadcopter to the left. The left motor should speed up. The right motor command should slow down. 3. Tilt the Quadcopter to the right. The right motor should speed up. The left motor should slow down. 4. Tilt the Quadcopter forward so that the front motor is lower than the back motor. The front motor should speed up. The rear motor should slow down. 5. Tilt the Quadcopter up (the front motor should be higher than the rear motor). The rear motor command should increase. The front motor command should decrease. 6. Rotate the Quadcopter clockwise. The front and rear motor commands should increase. 7. Rotate the Quadcopter counter-clockwise. The left and right motor commands should increase in value. The above Pre-Flight Tests were referenced from http://AeroQuad.info and modified for the Quadcopter. [22] There are also communication related tests that will be described further in Section 6.0 Wireless Communication.5.3.3 Flight Control and Results

As with many projects that rely on a variety of variables and have a large amount of components that need communicate with each other, this project encountered many issues during the integration period of the development cycle. This section will discuss the results and problems encountered for the flight components as well as the approach


51


April 7th, 2010

to solve them. Further discussion on integration

39369044 Internet Mobile Robot the Quad Copter

Documents

pulse width

inertial measurement

internet mobile

inertial measurement

nickel metal

de bothezat

full duplex

damping constant