Lightweight Fixed Wing UAV

Transcription

1 Lightweight Fixed Wing UAV Joseph Patton, Paul Owczarczyk, Mattias Dreger, Jason Bui, Cameron Lee, Cindy Xiao, Rijesh Augustine, Sheldon Marquis, Ryan Kapteyn, Nicholas Kwan Wong, Mark Pollock, Andrew Jowsey, Arshad Husain, Emmanuel Odeke, Emil Wyrod, Michael G. Lipsett, Duncan G. Elliott University of Alberta TEAM UAARG Abstract This report summarizes the design, build, and testing of a lightweight, versatile, fixed wing, unpiloted aerial vehicle (UAV). The UAV is comprised of a compact, customized expanded polypropylene airframe, a modular avionics board, an image capture system, and an electric power plant. The UAV flies autonomously and uses the avionics to perform real time image capture and transmission it also has an additional Wi Fi link for gaining access to remote networks. 1

2 INTRODUCTION The motivation for this project is to develop the next generation of unpiloted aerial vehicle (UAV). Current UAV systems are costly to replace and challenging to transport. As well, on board avionics are problematic to work on and debug because components are not readily accessible. To overcome these challenges, the goal was to develop a UAV that would reduce the cost, size, and complexity of the overall system while still meeting the requirements[1,2] of the AUVSI Seafarer Chapter 12th Annual Student UAS (SUAS) Competition. This system is a new system for the University of Alberta which was developed form the ground up for this competition. The system includes a lightweight, compact airframe housing a modular avionics board with an imaging system, autopilot, and telemetry radios and antennae, all powered by lithium polymer batteries. Figure 1 shows an high level overview of the entire system. The modular unit is designed with a simple D sub connect/disconnect wire harness so that hardware can be readily accessed, replaced, and/or debugged. While flying autonomously, the UAV navigates waypoints and searches an area of interest while the imaging system captures images, transmitting them to a ground control (GC) laptop, where image recognition software is applied to find objects of interest, such as colourful polygons and alphanumeric characters. Figure 1: System Block Diagram 2

3 SYSTEM OVERVIEW 1. AIRFRAME Figure 2: EPP FPV Airframe Inconceivable Table 1: Weights and Balances 3

4 The Hobbyking EPP FPV Airframe was chosen specifically because of its desirable features, characteristics, and low cost. The body is made of expanded polypropylene (EPP) and carbon fiber is small relative to the current UAV airframes has a large cargo area is designed to be a platform for image capture and is low cost. EPP foam, in contrast to other foams, has ideal characteristics that work well for UAVs: it is able to absorb impacts well with little to no damage or deformation. EPP foam, combined with carbon fiber, makes for a lightweight airframe relative to traditional remote control (RC) plane materials such as wood and plastic. EPP is also chemically inert, enabling a wider variety of construction methods. The compact dimensions (see Table 1) are an advantage as it aids in transportation and deployment of the UAV into remote locations. The EPP FPV costs only approximately $70 with shipping. This cost is even more significant when accounting for the consequences of crashes: EPP is quite durable, and an EPP aircraft could survive several more incidents than traditional airframe before having to be replaced. Another advantage to the Hobbyking EPP FPV is that it is designed for first person view (FPV) flight. It was developed with the intent that the operator would be taking pictures during flight. A significant portion of the project is image capture, it was an obvious choice of platform. Or test bench airframe was initially developed for use with a single battery, with the potential for later expansion, to reduce the overall mass of the aircraft. From initial estimates, the operational mass of this UAV was projected to be kg. It was important that the plane be versatile therefore it was important to guarantee that the structure is capable of future expansion. An overall mass of 3kg was used as a base line for a minimal flight requirement, making room for adding sub systems or batteries as needed. The test bench was flown at a mass of kg successfully by adding masses to the empty cargo area. The current system has a cargo space and avionics unit designed to accommodate two batteries in order to exceed the 40 minute flight time requirements. Table 1 shows all of the fully operational prototype weights and balances of with one battery. For preliminary testing purposes, it was only necessary to use one battery, with a flight time predicted to be minutes. Once all components were integrated into the airframe, accurate current draw and power drain values were established. These values determined the projected flight time of our batteries, as shown later in the power distribution discussion. Figure 3: Structural Modifications. From top left, clockwise: Payload Drop Apparatus, Body Adapter, Camera Mount, Motor Mount, Rudder Extension, Servo Mount, GPS Bay The air frame has specific customizations that were developed from initial concepts and subsequent testing. The out of box airframe is designed to carry smaller motor than the one currently incorporated. Therefore, it was necessary to reinforce the motor mount so that the added weight and larger components would not damage the airframe. The initial wood mount was removed and replaced with a more robust piece of 1/6 inch plywood. Long, aggressively threaded screws were embedded into the EPP as well to aid in adhering the mount to the fuselage. The motor attaches to four stainless steel bolts embedded into the 4

5 airframe by easily accessible nuts (Figure 3). In addition to reinforcement, this new system is more versatile, as it is a simple and quick way to remove the motor without degrading the material that it is attached to. The camera mount is designed so that the camera plugs directly into the wing (Figure 3). A wood mount was created that would attach directly to the wing via bolts for easy access and removal and the camera is held to the board with elastics. Finally, the rudder was extended to provide better performance and to improve handling (Figure 3). Air Drop Canister is a self contained modular unit that is embedded into the EPP foam on the belly of the plane. It is constructed of carbon fiber and epoxy, a building material that is well suited for UAV as it is strong and light weight. Its located relatively close to the CG so that any moment arm created by its added mass will be reduced. Because this is a pusher prop airframe it its also important to keep in mind the potential hazard for the ribbon to get into the prop. The current location of deployment is below the prop and therefor this hazard is mitigated. As well the unit is counter sunk so that belly landing EPP can still land without causing damage to the plane or the air drop unit. To activate the airdrop unit the Autopilot is programmed so that the gear switch on the transmitter can be operable in all modes of Autopilot so that a manual drop is always possible. The payload is immobilized by a 9 gram servo with a modified servo horn until the time of deployment. Once target deployment area has been reached the GCS operator will advise safety pilot to deploy payload at which point he pilot will engage the gear switch and the payload will drop from the plane. Deployment is determined by wind speed and direction as well as the speed, direction and altitude of the UAV. Conditions will be determined at the air field for optimal drop procedure and location. The location of the drop zone will be determined by flying over the target and obtaining the geolocation of the target, which will then be added in the GCS as a way point. The design goal of quick removal/replacement was carried over in all customizations. The GPS unit is embedded in the rear wing with a hatch for protection (Figure 3). It was determined that the CG was easily adjusted by the placement of the batteries. On our UAV, all tail wiring is run through the hollow carbon fiber shaft. The most significant addition to the new airframe is the body adapter (Figure 3). This adapter further reduces its size by making it possible to take the UAV apart right at the center. AIRFRAME TESTING Table 2: Flight Testing, Pilot Training Date Purpose Imaging Payload Drop Safety pilot Training Comment October 5, 2013 Maiden Flight no no no Crash, Rebuilt December 2,2013 3kg test no no no Initial Flight March 15,2014 RC config. no no no March 22, 2014 RC config. yes no yes April 4, 2014 Auto 1 Config. yes no yes April 26, 2014 Auto 1 Config. yes no yes May 2, 2014 Competition no no yes UAS Competition May 3, 2014 Competition yes no yes UAS Competition May 27, 2014 Payload Drop no yes no 5

6 Table 3: Payload Deployment Testing Test Distance (meters) Goal Result Deployment in manual 250 manual deploy Success Deployment in Auto Manual deploy with no AUTO interference Deployment in Auto Manual deploy with no AUTO interference Success Success Loss of RC 250 No deployment Success 2. MODULAR AVIONICS Within the UAV, there is a multitude of electronic components, which are all required to reside within the limited cargo space. Many of these parts also send and receive data wirelessly: these devices can thus easily interfere with each other, causing issues such as increased data packet loss, or even complete loss of connectivity. In order to keep interference to a minimum, the antennas for the wireless modules were distributed throughout the body of the plane. The distribution of weight is another main concern, as the CG of the entire plane is required to be near the leading edge of the wing. The table of weight distribution and the resultant CG is shown in Table 1. These components and their respective positions in the final design are shown in Figure 4. Figure 4: Avionics CAD Schematic Figure 5 shows the complete connection schematic of the components within the design. In this figure are 4 differently shaded sections representing the semi isolated electrical zones within our system. Near the bottom are the wireless modules for connections between the UAV and the ground station. The autopilot microcontroller, Lisa/M v2.0, was chosen for its inherent ability to connect to all of the servos, sensors, and transmitters, and for its native compatibility with the open source autopilot project Paparazzi. The ODROID ARM board is used for imaging and connection to a remote network. This computer was chosen for its small footprint, both in size and power consumption, as well as its ability to run Ubuntu. 6

7 Figure 5: Avionics Interconnections 3. POWER SYSTEMS The 4 cell 14.8 V lithium polymer (Li Po) battery, two DC DC converters, and a battery eliminator circuit embedded in the electronic speed controller (ESC) supply all the components requiring DC power. The ESC also supplies the motor with AC power. Figure 6 illustrates the power distribution system on the UAV. Efficiency tests were conducted on the motors. Table 2 shows a summary of the results. These results were based on the motor running at approximately half throttle. The voltages and current of all the components were measured wherever possible. Table 3 shows a summary of the results. The measurements confirmed that the power distribution of the system was sufficient and that no current or voltage ratings were exceeded. The endurance was calculated by taking into account the 5000mAh rating of the battery and measuring the current draw from the power supplies in the system. With the numbers shown in Table 2, the estimated endurance was 22.5 min. There are several different options to improve the design of the system. One suggestion is to implement a redundant power supply scheme for the flight controls. This would allow for safe landing in the case of a battery or power supply failure. The additional components required to implement this system would be a 2 cell 7.4V 500mAh Li Po battery, and two programmable voltage regulators that can be set to a level slightly lower than the main power supplies in the current system. These voltage regulators would only operate when there is a failure in the main system. The blue blocks in Figure 6 illustrate how the redundant power setup would be implemented. 7

8 Figure 6: Power Distribution Schematic Table 4: Power Plant Efficiency Tests Motor Propeller Input Voltage (V) Input Current (A) Input Power (W) Prop Speed (RPM) Output Power (W) Efficiency (%) Thrust (g) Eflite KV TurnigyG KV 10x7 APC x7 APC

9 Table 5: Voltage and Current Measurements Component Input (V) Output (V) Input (A) Output (A) Other 4S LiPo Battery mAh capacity ESC DC Supply DC Supply Odroid 5.03 Lisa/M Autopilot & Xbee 3.31 Primary Rx Secondary Rx GPS Servos Eflite 32 Motor COMMUNICATIONS The plane has four communication links: GPS, radio control, radio telemetry, and Wi Fi. These links are used to send information from the ground control station (GCS) to the plane, and to send information from the plane to the GCS. Figure 7 shows an overview of these links. The GPS unit is a stand alone receiver located at the tail of the plane. It was placed as far as possible from the other antennae to minimize the chance of interference. It uses an Ublox LEA H module operating at MHz, and has a refresh rate of 2Hz, providing an accurate location with minimized power draw. With 50 channels, the receiver is capable of acquiring a reasonably fast lock from a cold start. More importantly, it does not lose its lock if some GPS links are lost, and can establish a lock quickly if all are lost. This last feature is especially important because the plane's operation is dependent on knowing its exact location at all times. This Ublox module was paired with a ceramic antenna, and assembled into a GPS receiver by the Paparazzi community, and was designed specifically to work with the Paparazzi firmware. 9

10 Figure 7: Communication Link Overview Radio control is done using a 2.4 GHz receiver and transmitter. These are the same transmitter and receiver that were used in previous years; therefore, the transmitter is tried and tested. The receiver and transmitter use DSMX as a communication protocol. This protocol is similar to the common DSM2, but adds channel hopping on 24 different channels. This feature increases the range and reduces the possibility of interference in the crowded 2.4 GHz band. The radio control link (RC) is used mainly during manual flight, letting the user control the plane directly by line of sight. Once the plane enters autopilot mode Stabilized Flight (AUTO1), the user still has control of all flight axes, but is restricted in range by the autopilot configuration. This permits the user to control the plane via the GCS flight display without requiring line of sight, and the autopilot prevents the plane from rolling over or crashing. In fully Autonomous Mode (AUTO2), the autopilot is in complete control of the plane (with the exception of the rudder, which is still pilot controllable). The transmitter is also used to control the egg drop servo, while in manual and in any of the two AUTO modes. The radio telemetry link is responsible for transmitting data between the plane and GCS. It is one of the most critical links since it transmits to the ground station, and the user, the position and state of the plane. For this reason, the 900MHz XBee PROs are used to create the link. They were chosen specifically for their long range and operating frequency, which is different from the other antennas on board, further reducing the chance of interference. The GCS and avionics XBees are paired by matching the destination addresses on both devices. This increases their reliability and reduces data packet loss. Since transferring image data requires a large throughput, the plane uses a Wi Fi network to transmit pictures. The plane has two on board Wi Fi antennas. One is used for transferring the images and to wireless access the onboard imagine computer, while the other is used to either, search, and retrieve files from foreign networks, or to bridge a Wi Fi network and act as a relay. The link operates at 2.4 GHz and a local area network is established using a router. The standard antennas on the router are replaced with a large, high gain, directional antenna. This maximizes the range of the network but reduces its beam width; therefore the antenna needs to be aimed toward the plane during flight. The planes built in Wi Fi antennas access this network automatically and begin sending down images along with location data for processing on the GCS. When the Wi Fi link is lost, the images are stored onboard and restart transmitting once the link is reestablished. The loss of any of these radio links while flying is a flight safety concern. To reduce these hazards, the plane is programmed with a procedure for reacting to a lost link. In manual or Auto1 mode, the loss of any of the data links has no effect on the plane; however, the loss of the RC link means that the pilot loses control of the plane. If this link is not re established after 4 seconds, the plane will enter a near vertical dive and will terminate the flight. If the radio telemetry is lost in Auto2 mode, the plane will wait for a reconnection, and if after 10 seconds the link is still down, the plane is programmed to fly to and then circle above the GCS. If after 2 minutes this link is still down or the pilot has not returned to manual mode, then the plane will terminate its flight. If the GPS link is lost, the pilot must return the plane to manual mode; if this is not an option, then the plane will terminate the mission. The RC, GPS, and telemetry links were tested by establishing a GCS on a flat stretch of highway and testing the throughput, signal strength at different ranges, and antenna orientation. From these tests it was determined that the RC transmitter has a reliable rage of 2.5km and reliable telemetry up to a distance of 2.75km. These numbers are used as our maximum operating distance, since operating outside this range results in lost links and unreliable data throughput. 10

11 5. AUTOPILOT Paparazzi is an open source system of hardware and software used to create an autopilot system. Additionally, it is constantly developed by hobbyists and researchers around the world. Unlike other commercial autopilot systems, Paparazzi can be modified and altered to suit a variety of applications. Therefore, this system has proven to be a versatile and reliable system for UAARG s use. There are two separate components to the Paparazzi software: the ground control station (GCS) and on board firmware. The on board firmware is flashed to a microcontroller, the Lisa/M v2.0, which is integrated to control all of the servos and the motor. The IMU detects the aircraft s speed, orientation, and all relevant axial data. With this information, the microcontroller changes the motors and servos in order to correct any deviations from the pre programmed flight path. It does so by using proportional integral derivative (PID) control loops implemented in software, with gain constants set through calibration via flight testing. If appropriately tuned, the autopilot firmware will stabilize accordingly to disturbances such as wind, changes in course, etc. At the GCS, Paparazzi provides a graphical user interface. During testing and preparation, this software is used to run simulations on sensor data, adjust looping gain constants, and to set UAV flight plans. Mid flight, the GCS is primarily used to monitor aircraft s location, attitude, and telemetry link strength. As an additional feature, the flight plan can be adjusted as needed to accommodate for changing mission objectives. The Lisa/M v2.0 is the second generation of the Lisa/M Paparazzi microcontroller. This microcontroller is highly adaptable to most aircrafts, lacks few additional components, and its documentation is always being updated. The controller has an attached inertial measurement unit (IMU) on its bottom side. The Aspirin version 2.2 is soldered directly to the bottom side of the Lisa/M and communicates using a I2C port. This crucial piece of hardware includes a 3 axis accelerometer, 3 axis gyroscope, 3 axis magnetometer, and a barometer. With these sensors, the autopilot firmware can measure the UAV's exact acceleration, orientation, and heading. The pre flight setup is further simplified as the accelerometer, gyroscope, and barometer are pre calibrated. However, the magnetometer required calibration to compensate for the magnetic fields created around the avionics board. Without compensating for these relatively powerful magnetic fields, which completely drown out the Earth's weak magnetic field, the magnetometer would be useless. The user controls the function of the controller through a set of XML files: an airframe file, mission file, and telemetry file. When the firmware is flashed, all 3 files are parsed and assembled into the main autopilot code. The airframe file describes the: plane's physical dimensions, number of servos, control inputs, servo travel, connected hardware, and the orientation of the IMU on the avionics board. It also defines the control loops used, and the corresponding control loop gains. The mission file gives details about the mission, such as the location of the ground station, boundaries, and objectives. The telemetry file lists the data format along with the frequency that is sent down the telemetry link. The gains in the control loops need to be fine tuned for reliable autonomous flight. If these gains are not tuned from their default values, the autopilot s attempts to stabilize could be catastrophic. Tuning is an extensive process that requires hours of flight time and simulating. The procedure involves isolating each control loop and tuning it individually. Since the loops are in a hierarchy, the basic loops are tuned initially while more complex loops are adjusted later. An overview of the tuning process follows: First, the aircraft is trimmed and adjusted mechanically so that it is capable of stable flight under manual control; without requiring the transmitter to be trimmed. Next, the plane is flown in a series of pre determined maneuvers with a data logger running. When the log files are graphed, values such as maximum roll, pitch, and throttle can be extracted and placed in the airframe file. At this point, the plane should be stable enough to enter Stabilized Flight (AUTO1) mode without loss of control. In AUTO1, the plane is flown in a series of maneuvers to test its responsiveness. If they are too slow, then the gains are increased; if the plane oscillates, then the gains are reduced. The goal is to adjust gains to get the most responsive control with no oscillations. Once AUTO1 is capable of keeping the plane stable through pitch and roll loops, the plane will be stable enough to activate fully autonomous flight (AUTO2). In the following two graphs, a plot between the autopilot s desired attitude and in flight attitude is given. With the gain constants nearly complete, the parallel rises and falls provide insight into the PID system. Figure 9: Desired Pitch vs. Actual Pitch 11

12 Figure 10: Desired Roll vs. Actual Roll 6. IMAGING AND WI FI NETWORKING The imaging and networking subsystem is responsible for capturing images and image data from the aircraft, relaying it to the ground, and processing that data to obtain target information. There are two major components to the system: The onboard payload hardware and software the ODROID U2 onboard imaging computer, ODROID USB Wi Fi interface cards, the Point Grey Chameleon still imaging camera, and custom written software, Waldo, for capturing image and image data and transmitting it back to the ground station. The ground station the Ground Station Router for establishing a Wi fi link with the aircraft, the Ground Station Laptop for receiving the image data, and custom written software, Pigeon, for extracting target information from images. The on board components are responsible for image capture, remote network access, and relaying data to the GCS laptop. The GCS laptop in turn is responsible for image processing/analysis and to act as an interface to the user. The GCS router establishes a local Wi Fi network to enable communications between the UAV and the GCS laptop. The main board used on the UAV is the ARM based ODROID U2, running a stripped down version of the Ubuntu operating system. It is connected via USB to 1 ODROID USB Wi fi interface card to allow the ODROID to connect to wireless networks. Image data is obtained using the Chameleon CMLN 13S2C CS still imaging camera. Information about the aircraft s 12

13 position and attitude is obtained from the Lisa/M autopilot board and transferred to the ODROID via a UART to USB adapter. This hardware layout can be seen in Figure 9. Figure 9. Imaging system overview The software running on the imaging computer, Waldo, controls and receives images from the camera, receives position and orientation data from the autopilot board, and combines those together so that each image captured has information about the plane s behaviour at that time. The imaging software uses the library libdc1394 to control the Chameleon camera, gathering images at a configurable rate and temporarily saving them locally as JPEGs. The software also listens in directly on raw binary data from the autopilot via a UART to USB interface; GPS and altitude data are filtered out from this data stream, and are used to geo tag images. The data is then transferred to the ground station laptop via the FTP protocol. The image capture, processing, and transmission tasks are each captured into their own thread in order to allow the tasks to be run efficiently concurrently. Waldo is initiated remotely from the ground station laptop, and the user is able to monitor the status of image processing and transmission from the ground station via an SSH connection. After initiation, Waldo runs regardless of whether a Wi fi connection exists between the aircraft and the ground station; it saves images into a local buffer when link is lost and attempts to re transmit images upon regaining a Wi fi link, ensuring its reliability. Transmission is done by uploading files to the FTP server that runs on the ground station. In order to develop a strategy for covering the search area, we have performed calculations that take into account the following properties of our system and the search area environment: The effect of different lenses on the Point Grey Chameleon camera The effect of camera shutter speed The size of the search area The amount overlap desired on each image These are needed to calculate properties that we may control. These include: Rate of image capture (represented by Seconds per Image in the spreadsheet), configurable via Waldo The nominal velocity and altitude for flight, configurable via Paparazzi The amount of mission time to spend inside the search area 13

14 A spreadsheet with an example calculation is shown below, for a 12.5mm lens and 1/60s shutter speed. This shows that we are able to cover a 1km x 0.5km search area comfortably within 18 minutes and ~500 pictures, even with a 10% overlap for eachimage. Table 6. Example search area calculations At the ground station, the images are processed using a custom written ground station software, Pigeon, that presents the images in a gallery to ground station operators. Pigeon is primarily intended to allow the ground station operators to find targets in the images received. Once a target is found in an image, it is selected by the operator and its properties (shape, colour, etc.) are entered and saved by the operator. The plane s position and orientation data are used to find the true position of the targets an algorithm has been written to correct for the plane s pitch, its roll, and position of the target selected in the image. Pigeon is also able to export the image data to Google Earth, in order to provide an overview of the flight path of the plane. 14

15 Fig. 10. Ground station Pigeon image analysis program. Test image shown, with visible 4 x4 square blue target with 2 wide red letter T oriented north Preliminary georeferencing results using this system are shown below. Table 7. April 30th georeferenced data results Image # Target 110 Red T on blue Black X on 231 white Vertical and horizontal 383 stripes 386 White R on red 6x6 387 Checkerboard 4x4 539 Checkerboard Vertical and horizontal 540 stripes Stripes and 556 pinwheel 608 Red T on blue Reference Landmark Actual Location , , , , , , , , , Georeferenced Location Discrepancy (m) Discrepancy (ft) , , , , , , , , ,

16 7. ACHIEVABLE COMPETITION OBJECTIVES a. AUTONOMOUS FLIGHT In this competition, UAARG will be capable of performing fully autonomous flight. By using a reliable autopilot system in conjunction with numerous flight tests and tuning, the aircraft will preform stabilized flights between waypoints and sectors. Furthermore, the autopilot will respond appropriately to unpredictable scenarios such as strong winds, loss of RC link and loss of telemetry link. With all links and systems intact, the aircraft will guide its way home after the necessary images have been acquired. b. WAYPOINT NAVIGATION Prior to launch, a flight plan will be created in a XML file for the Paparazzi autopilot system to use. This flight plan will include waypoints and sectors for which the aircraft is designated to fly. As it navigates through the airfield, the users and judges will be provided information regarding its position (overlaid on Google Earth images), speed and attitude. Images will be taken continuously as the aircraft flies above the waypoints and search areas. During and after the flight, photos will sent to ground station via wifi connection. At ground station, images will be analyzed and documented. Using UAARG s geo referencing software, details such as target color, shape, orientation and location will be determined during and after flight time. Once all targets have been recognized, ground station members will proceed deciphering the secret message hidden within the alphanumeric symbols. c. ACTIONABLE INTELLIGENCE As the aircraft gains valuable images, it is important to deliver results as soon as possible. To do so, a wifi connection for image transfer must be established between the ground station and the flying aircraft. To improve connection reliability and strength, a yagi antenna is used to broadcast the server from ground. This allows the aircraft to venture further distances while maintaining communication. Furthermore, if power failure or interference occurs, the on board Odroid is programmed to automatically reestablish connection and to continued exporting images. By having pictures available to ground station, analysis can begin before landing. As a result, target characteristics are discovered well in advance to mission deadlines. d. OFF AXIS TARGET To capture and analyze an image that does lie directly beneath the aircraft, it requires communication and integration between autopilot and imaging. By allowing the imaging software to tap the attitude and GPS data from autopilot, geo referencing is achieved through an algorithm designed by UAARG. In the flight plan, waypoints are set to notify the autopilot when to perform a banking roll. With the aircraft is angled, images of unreachable waypoints are taken. In conjunction with the algorithm, details such as target color, shape, orientation and location may be determined. e. EMERGENT TARGET TASK Due to the autopilot s flexibility, waypoints can be changed mid flight using the GCS. Once the new waypoint is recognized, ground station members will add the coordinates for the aircraft. The aircraft will fly to the location and proceed taking sufficient photos for determining the target s color, shape, orientation and location. f. SIMULATE REMOTE INFORMATION TASK (SRIC) Currently, the hardware of avionics board is capable for integrating a second wifi card into the system s Odroid. This second wifi card would provide the necessary access between the aircraft and the waypoint. Although the hardware is capable, the software is not prepared to obtain data from a remote destination. g. INTEROPERABILITY TASK Due to open source nature of Paparazzi software, GPS data acquisition is simple. During flight, the autopilot is continually absorbing, analyzing, and documenting data such as location and attitude. By tapping serial data, the aircraft's GPS coordinates are streamed to the ground station via wifi connection. Sequentially, the data is then imported in to Google Earth. As a result, spectators are provided a live stream of the aircraft s whereabouts in relation to satellite imagery. 16

17 h. AIR DROP CANISTER In order to demonstrate UAARG s ability to successfully provide emergency rescue canisters, the following conditions are to be met: the aircraft is less than 200ft away from the target, aircraft altitude is between 300ft and 400ft, and airspeed is greater than 25kts. In this drop mechanism, a flour filled plastic egg shell (not exceeding 2 in length and 1.5 in diameter) and a connected ribbon (5 long and 1.5 wide) will be held in a self contained module installed on the bottom of the EPP airframe. To avoid compromising the structure or the weight of this airframe, the module s material is comprised of carbon fiber sheets. To ensure that aircraft may still preform a belly slide after landing, the module was counter sunk within the fuselage; Thus, preventing any items to become caught during the process. A controlled release is accomplished through a 9 gram servo installed orthogonal to egg shell. By rotating the servo 90 degrees, the canister falls out from the force of gravity. By modifying the airframe file, stored in the autopilot s firmware, an RC control option can be added. This allows the pilot to release the canister by simply flipping a switch located on the transmitter. 8. CONCLUSION The University of Albert Aerial Robotics Group set out to design, build, test and deploy an unmanned aerial system. We began this year with a completely new air frame and strategy. Our goal with the new strategy was to solve the challenges of previous years with a complete re engineering of the UAV and subsystems. We built the air frame and test benched all subsystems before ultimately doing integration. Through testing and tuning on the UAV we are confident that we will accomplish many of the mission requirements. The design and construction of the UAV was successful as the client's requirements were achieved. The selection and subsequent modification of the airframe resulted in an excellent platform to run all the subsystems on. The power systems design provided the currents and voltages needed to operate all subsystems. The avionics board was compact and easy to debug. The selection and integration of electronic components and antennae performed as required. There are still areas of improvement that can be addressed. These include: modification of airframe surfaces for better aerodynamics; further clean up of wire clutter in the avionics; the implementation of a two battery avionics board; improved wireless communications throughput; and further autopilot refinement to achieve complete autonomy. 9. ACKNOWLEDGEMENTS We would like to thank Peter Cary, the Edmonton Radio Control Society, the University of Alberta, Shell Canada and all members of the University of Alberta Aerial Robotics Group as for their support and guidance. 10. REFERENCES [1] Unmanned Systems Canada. 6 th Unmanned Systems Canada Student UAS Competition Internet: Nov. 16, 2013 [Apr. 6, 2014] [2] AUVSI Seafarer Chapter. AUVSI Seafarer Chapter 12 th Annual Student UAS Competition Internet: seafarer.org/documents/2014documents/2014_auvsi_suas_rules_rev_1.0_final_13_1024_1.pdf, Oct. 24, 2013 [Apr. 6, 2014] 17