American Institute of Aeronautics and Astronautics

1

Benefits of Autopilot Integration for Enhanced UAS Operations

Mariusz Czarnomski1, Richie Spitsberg.2, David Dvorak3, Richard R Schultz4 and William H. Semke5

Unmanned Aircraft System Engineering (UASE) Laboratory, School of Engineering and Mines, University of North Dakota, Grand Forks, ND 58202, www.und.edu

Abstract To enhance Unmanned Aircraft Systems (UAS) operation of a small UAS vehicle an

autopilot was integrated into the aircraft. The integration allowed the UAS Engineering Laboratory team to fully understand the operation and programming of the autopilot to better utilize its capabilities in a systems engineering approach. The integration process gave the team valuable experience and understanding of the autopilot system, which could later be used in the field to troubleshoot future problems without the need of interrupting a flight day. The payloads, aircraft, and autopilot act as a system rather than as independent components to accomplish specialized missions. This requires a detailed and thorough practical knowledge of the autopilot system to fully exploit its capabilities. Through the knowledge gained, we were able to combine data from the autopilot with payload data to utilize the complete UAS system more efficiently.

Nomenclature ADS-B = Automatic Dependent Surveillance-Broadcasting AVL = Athena Vortex Lattice BTE = Bruce Tharpe Engineering CCD = Charge Coupled Device CG = Center of Gravity COA = Certificate of Authorization EMI = Electro Magnetic Interference FAA = Federal Aviation Administration EO = Electro Optical GPS = Global Positioning System IAS = Indicated Air Speed IR = Infrared LiPo = Lithium-Polymer LUT = Look-up Table NAS = National Airspace System NiCad = Nickel Cadmium POH = Pilot Operating Handbook RC = Radio Controlled RF = Radio Frequency UASE = Unmanned Aircraft System Engineering UAS = Unmanned Aircraft System UAV = Unmanned Aircraft Vehicle

1 Undergraduate Research Assistant, Dept. of Electrical Engineering, 243Centennial Dr. Stop 7165. Student Member 2 Graduate Research Assistant, Dept. of Mechanical Engineering, 243 Centennial Dr. Stop 8359. Student Member 3 Undergraduate Research Assistant, Dept. of Mechanical Engineering, 243 Centennial Dr. Stop 8359. Student Member 4 Professor/Chair of Electrical Engineering Dept., 243Centennial Dr. Stop 7165. Not a Member 5 Associate Professor of Mechanical Engineering Dept., 243 Centennial Dr. Stop 8359.

AIAA Infotech@Aerospace Conference <br>and <br>AIAA Unmanned...Unlimited Conference 6 - 9 April 2009, Seattle, Washington

AIAA 2009-1995

Copyright © 2009 by Mariusz Czarnomski and Richie Spitsberg. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.

American Institute of Aeronautics and Astronautics

2

I. Introduction The main objective of the project was to successfully integrate the Cloud Cap Technology Piccolo II autopilot

system into the BTE Super Hauler aircraft, followed by autonomous flights with different payloads integrated onboard. Data collection was an integral part of each payload flight, and is divided into two groups; payload data (application specific) and data from onboard avionics (autopilot) to examine full system performance. Both sets of data were closely analyzed after each flight. The data from the payloads is used for more specific applications, and it is also used to improve the payload system operation (i.e., tracking capabilities of a gimbal system, post flight data analysis, and the generation of video mosaics). The data retrieved from the autopilot was used in two different ways. It was used in combination with the payload data and served as a reference point comparison, or simply as a source of flight dynamics information. For example, the GPS, attitude and altitude data from the autopilot are directly used by our ADS-B payload for the sense and avoid capabilities of the system. Having the latter data on hand combined with the data from ADS-B unit, the ADS-B payload can predict possible collision with other aircraft, and consequently order the autopilot to perform collision avoidance maneuver. The data was also used to analyze the autopilot performance, fine-tune the system, and troubleshoot any problems encountered. The benefits and advantages gained from the process of autopilot integration into the UAV cannot be emphasized enough and were kept in mind throughout the development of the system. Students and their advisors truly believed that this complicated course of action was superior to the alternative of acquiring a UAV with an autopilot system already installed. During the integration process, students were able to become more familiar with the autopilot itself, including its design and capabilities, such as post flight data analysis, to gain a better understanding of the UAS system as a whole [1-8]. Another crucial point is that the payloads are an integrated part of the UAS, and they can directly interface with the autopilot. This operation requires not only knowledge of the autopilot operation and interface procedures, but also demands modification of the aircraft itself. All tasks were successfully achieved thanks to the opportunity of the autopilot integration process.

II. Steps in the Process of Integration The entire enterprise was a multi-step process that required a systems engineering approach, which began with

the collection of crucial parts and components. The list consists of the aircraft, autopilot, ground station, GPS, 900 MHz antennas, and cabling, as shown in Figure 1. A custom designed aircraft was purchased from BTE due to the specific requirements regarding the material used for the airframe and the size and shape of the payload bay. The aircraft needed to be easily modifiable, have a rectangular payload bay, and be capable of carrying payloads of up to 30lb. The choice for the autopilot was the Piccolo II by Cloud Cap Technology. This decision was based on reliability and ease of compatibility with other small UAS systems that our team has worked with in the past (e.g., Lockheed Martin and Raytheon Company). Following that decision, many items were purchased, including a pitot/static tube for air data collection, vibration mount for the Piccolo II, deadman/tachometer board for engine data collection and flight termination in case of system failure, servo wiring harness, and an appropriate ignition system that could be interfaced with the deadman/tachometer board. Two Piccolo II autopilot units were purchased, so that one could be installed in the aircraft and the other could be used for lab testing. The second unit also served as a spare in case the other failed. A CCD camera with transmitter and antenna was purchased for forward looking capabilities. Additionally, there were five antennas mounted on the aircraft to be used for payload configuration, autopilot communication, and GPS navigation. The autopilot, forward looking camera, and payloads are powered by LiPo batteries, with the aircraft’s servos and ignition being powered by NiCad batteries. To ensure safe operation of the entire system, two sets of batteries were purchased and kept charged at all times. Before and after each flight, the health of all batteries was checked to guarantee successful and safe flights, and if need be, the batteries were replaced. Furthermore, the operator of the autopilot can constantly monitor the voltage of the autopilot and servo batteries from the autopilot console.

After collection of the critical components, the next step in operation was to become familiarized with the Piccolo II autopilot system, which was done by studying the available Piccolo II documentation that came in the autopilot package. In the documentation, outline of the integration process could be found. The documentation provided does not lead the user step-by-step through the integration process, but rather helps direct him along the correct path.

American Institute of Aeronautics and Astronautics

3

The integration process is system-specific, and it would be virtually impossible to have installation instructions that would work for every UAS configuration. In addition to the usual documentation, Cloud Cap offers an online forum where customers can ask questions regarding any product. A response to your question is usually posted within two hours, and if the problem requires more specific and in-depth analysis, Cloud Cap engineers contact the customer personally either by phone or via e-mail. We must say that Cloud Cap offers one of the best customer service experience that we have ever encountered, and this was another reason why we chose their product for the development of our system.

One of the initial steps Cloud Cap recommends is the development of a computer model of the aircraft so that it can be flown in the Piccolo II simulation. The model created becomes your only link between simulation and actual flight. This step, often omitted by others, was strongly emphasized, and requires a model of the aircraft to be developed as accurately as possible. This model requires the dimensional measurements of the control surfaces, wings, tail section, fuselage, landing gear location, and the location of the center of gravity. An accurate measurement of the location of the CG is critical, as most measurements are in reference to the CG location.

After the detailed physical information of the aircraft was collected and recorded, a computer model of the aircraft was created using a program called AVL, supplied by Cloud Cap in the software package [9]. The AVL runs a virtual wind tunnel over the model and develops its aerodynamic coefficients that are later used by the autopilot to

Figure 1: UAS autopilot system components (courtesy of Cloud Cap Technology www.cloudcaptech.com )

control the aircraft, as shown in Figure 2. After the simulation is run, the aerodynamic coefficients are saved in an XML file generated by the AVL. The AVL program is not that complicated of a program, but having no knowledge of it and having to learn everything from scratch proved to be difficult. Because of this much of our knowledge came from trial and error and support from the engineers at Cloud Cap. The next step was to create a LUT for the propeller and the engine. To create the LUT for the propeller, two other programs supplied by Cloud Cap, Prop.exe and JavaProp, were used. The necessary parameters required for input are the pitch and diameter of the propeller, expected RPM, spinner or prop hub diameter, and design airspeed. The programs create a LUT containing thrust and torque data of the propeller based on the RPM and forward speed of the aircraft. Next, the simulator calculates the advanced ratio, and based on the LUT, determines the coefficient of thrust and coefficient of power that are used to calculate the thrust of the propeller.

Figure 2: Screenshot from the AVL program

The data required for the engine LUT was not provided by the manufacturer, causing this table to be more of a challenge to create. The required information for the table was the maximum horsepower, minimum and maximum RPM, and power output at specific RPMs. Since no information for the engine was available, an engine with similar specifications and data was found. Based on the data that was collected from the other engine, an estimate of our motor data was extrapolated into an LUT. Finally, the information from the XML file, and the propeller and the engine LUTs along with the aircraft weight, CG location, and landing gear location were compiled into a .txt file which were used directly by the Piccolo II during both simulation and actual flight.

After the .txt file was created, flight simulations using the Piccolo II software were conducted. In simulation mode, the aircraft behavior was able to be tested as if it was an actual flight. If the model in the simulation did not perform as expected, necessary corrections were made to the autopilot gains without the need for field testing the aircraft and possibly risking damage. The external pilot for operations was able to take manual control while in simulation and fly the model around to see if it responded and behaved properly. After a few iterations of gain adjustments the simulation model flew to the liking of the external pilot and was cleared for actual flight. The benefits of this simulation cannot be emphasized enough, since it allowed the pilots to see and feel how the aircraft would perform, in the safety of a lab environment and without even firing the engine. The performance of the aircraft in simulation is directly related to the accuracy of the model and measurements. Since such precise measurements were taken and no steps were skipped, the team was quite confident that the aircraft would perform the way it should when the autopilot was turned on for the first time during actual flight.

III. Hardware Integration Hardware integration into the airframe incorporated many different challenges into our work. Integration of the

Piccolo II autopilot system, deadman/tachometer board, and servo harness provided difficulties with EMI concerns requiring different methods of shielding to be included in the integration of the components. EMI issues occurred all throughout the integration process and were dealt with the best we could. The main concern was that the EMI would cause the servos to operate incorrectly during flight, resulting in damage to the aircraft. Copper tape was used to

American Institute of Aeronautics and Astronautics

4

cover all of the servos and to make ground planes for the antennas that were causing noise in the system, as seen in Figure 4. Upon completion of the shielding the servos operated much better and we were confident that operations would be successful.

Figure 4. Copper tape shielding and antenna ground plates From left: GPS Antenna, Servo, and Piccolo II antenna

Many of the wires that ran on the inside of the aircraft, including all the servo and ignition cables, were rewired and shielded with an EMI safe casing, as seen in Figure 5. Servo “chatter” due to interference was something that was studied for many hours and never completely solved and even with the precautions that were taken, there was still a residual amount of chatter in the servos.

Figure 5. Servo harness and ignition wire shielding

Throughout integration, many tests had to be conducted to gain the necessary data required for operation of the

autopilot. Control surface calibration and gain adjustment data was collected in order to simulate the flight of the aircraft with and without the autopilot system enabled. The angle of deflection of the ailerons, elevator, rudder, flaps, and throttle required accurate measurement in order to set the optimal gain levels for the control surfaces. The collected data enabled a simulation to be executed to observe how the aircraft would perform under both manual and autopilot control. From the performance of the simulation, adjustments were made to optimize the flight characteristics of the system.

A significant number of engine tests were conducted after the autopilot had been fully installed. The tests were conducted to observe whether or not the noise and vibrations of the engine were going to have any effect on the autopilot system. It was necessary to gain valid information from the engine due to the system requiring the engine data for airspeed calculations, which is calculated using both the IAS and the RPM of the engine. Engine testing was completed to establish whether or not feedback from the ignition and tachometer board received by the autopilot was correct. Using an optical handheld tachometer, the RPM was measured and compared to the data displayed on the autopilot console.

American Institute of Aeronautics and Astronautics

5

Aside from these tests, extensive system testing was conducted on the ground, including communication performance, battery performance, and testing with a variety of different payloads. The communication tests were conducted outdoors near the lab area. The plane was taxied under manual control to ensure that there would be no communication loss. Another range test was performed at the distance of a football field, with the aircraft being taxied on the grass. Both of the tests were successful. These tests were initially completed with no payloads in the aircraft and again with the different payloads that have been developed in the UASE Lab. Since there have been a multitude of modifications done to the aircraft, a large number of batteries is required to run the entire system. Since there are a number of systems that need to be running at all times, knowing exactly how long the batteries will last is a necessity. Battery life tests were conducted on all of the batteries under full operation of the system to ensure that the shortest time the batteries can run until they become a problem to the system was known. As a precaution, the critical batteries are checked after any extended flight and replaced if necessary.

A number of payloads have been developed in the UASE Lab, ranging in complexity from a simple digital camera to a multi-camera system controlled by a three axis gimbal. With the primary research on campus being payload development, it was important for the aircraft to respond the same for any of the payloads in the aircraft. The autopilot is mounted directly behind the payload bay, so there was some initial concern regarding EMI complications between the payload and autopilot. The payload bay was heavily modified by lining the entire bay with copper tape and placing aluminum plates on the front and back walls. All wires coming into the bay were shielded and passed through a small hole lined with EMI gasketing, as shown in Figure 6.

Figure 6. Payload bay and aluminum firewall with EMI shielded pass-through hole.

IV. Standardization Manuals Our UASE team has been working in conjunction with the John D. Odegard School of Aerospace Sciences at

UND in obtaining COA from the FAA for the testing of our small UAS in the NAS. It has been crucial that the guidelines provided by the FAA for normal aircraft be followed. Throughout our research and development, a fully detailed maintenance manual and a POH have been developed to fulfill FAA airworthiness requirements. Pre-flight checklists for the autopilot, aircraft systems, and payloads have all been developed, as well as a standardization manual for flight operations. The preflight checklists and standardization manuals are not required for FAA airworthiness, but they are required by UND for operation of the aircraft. The aircraft being operated is owned and insured by the university, so no matter where flights occur, their rules must be followed. In addition, after each flight, a report has to be filed indicating the number of flights, time spent in the airspace, and any problems encountered during the flight mission. Here the knowledge of post flight data analysis from the autopilot came in handy, and allowed for meticulous flight analysis.

V. Problems Encounter During Integration

Many problems were encountered throughout the autopilot integration process including; EMI and RF interference, ignition system integration, and aircraft modeling as mentioned earlier. EMI and RF interference were

American Institute of Aeronautics and Astronautics

6

American Institute of Aeronautics and Astronautics

7

two major concerns that were encountered throughout the integration process due to the multitude of critical frequencies that were being used by the different devices onboard the aircraft. Every electrical component on the inside and outside of the aircraft had to be tested for interference, and in most cases shielded for protection against EMI in order to not interfere with each other during operation.

The ignition system was a critical aspect of the autopilot system in the way that it was necessary for the operation of autonomous flight. Tachometer capabilities are required for power estimations used by the autopilot for air speed adjustments, which the initial ignition system did not have, thus requiring us to purchase a new system and interface it into the aircraft. The new ignition system is made by the same company that made the original system, but has the addition of the tachometer output.

VI. Results

In mid-April of 2008 the UASE Lab team of undergraduate and graduate students from the University of North Dakota received their first UAV. The aircraft is a BTE Super Hauler that supports UASE payload operations and development. Previously, students were dependent on the availability of other UAV operators, namely Lockheed Martin and Raytheon Company, to fly and test their payloads. The cooperation with these two companies was a great experience, but an increasing number of payloads developed in the lab and their functionality called for an in-house airframe capable of performing autonomous test flights.

The first flight for the UASE team took place on July 29, 2008, with subsequent flights on July 31 and August 1 at the Camp Grafton South National Guard facility near Devils Lake, ND. The reason for flying at the military base was the need for restricted airspace, since UAVs are currently not able to fly in the NAS without the correct paperwork. What was learned from this experience is that a major part of a successful flight day are well prepared and planned logistics.

Upon arrival, the ground station for operation of the autopilot and various payloads were setup inside a military tent, as seen in Figure 7, and all the system components were tested and prepared for flight.

With only three days to fly, it was decided that the first day would be designated to only RC flight, to allow the external pilot to get comfortable with the operation of the aircraft. During the pre-flight procedures, some of the sensors readings coming from the autopilot were incorrect and could not be fixed at the field. This problem provided the opportunity to use the spare autopilot. Day Two started with a couple RC flights followed by autonomous flights without payloads. A “dummy” weight payload was then flown in the aircraft to see how it would perform with the added weight. The results were satisfactory, with the aircraft actually flying more smoothly with the added

Figure 7. Ground station set up inside the military tent

weight. The rest of Day Two, and also Day Three, were dedicated to flying the various payloads, shown in Figure 8, that have been developed over the past three years.

Some major problems that were encountered during autonomous flights were inaccurate tracking of the flight path and sudden dives while switching between manual and autopilot control. The telemetry files from the flight were sent to Cloud Cap for a more intensive look at what could have been causing these problems. It was determined that the roll-rate-to-aileron proportional gain and integral gain values were set to high. These problems were corrected in the lab by gain adjustments on the specified control surfaces, and also aileron effectiveness.

American Institute of Aeronautics and Astronautics

8

Figure 8. Payload installation and Super Hauler flight

From the success of the first autonomous flight, multiple payloads developed by the UASE Lab team have been flown for data collection. A total of three different payloads were flown during the flight tests, including an ADS-B payload, an EO/IR 3-axis gimbal, and a multi-spectral camera payload used for precision agriculture purposes. The data collected from the three different payloads is being used in many different post-processing applications for further development of each of the respective systems.

VII. Conclusions Although our first autonomous test was a success, that is not to say that all the flights afterwards were complete

successes. Many different problems occurred throughout our first test week, and they were dealt with accordingly. Some of the problems included a faulty gyro in the autopilot unit, weight and balance, battery failure, payload-ground communication, and landing gear failure. These problems can often be solved in the field but sometimes they require a much more extensive examination of the system to assess what went wrong.

In the end, an autopilot was successfully integrated into a small UAS and its enhanced capabilities have been

utilized by several payloads and flight data analysis. This work currently is and will continue to be instrumental in the continued development of sophisticated payloads developed for surveillance, reconnaissance, as well as detect sense and avoid by the UASE team.

American Institute of Aeronautics and Astronautics

9

The key in the success of our missions and operation of the UASE Lab team is good communication between everyone that is involved with the project. Learning from your own mistakes might not be the most efficient way of doing things but it is effective in a long run. Nevertheless, at University of North Dakota, we try to do our job as adequate and error free as possible to prepare ourselves for future careers in a variety of engineering fields.

Acknowledgments This research was supported in part by Department of Defense contract number FA4861-06-C-C006,

“Unmanned Aerial System Remote Sense and Avoid System and Advanced Payload Analysis and Investigation”, and the North Dakota Department of Commerce, “UND Center of Excellence for UAV and Simulation Applications.” The authors would also like to acknowledge the contributions of the UASE team at UND.

References

1Ranganathan, J., and Semke, W., “Three-Axis Gimbal Surveillance Algorithms for Use in Small UAS.” Proceedings of the ASME International Mechanical Engineering Conference and Exposition, IMECE2008-67667, November, 2008.

2Semke, W., Schultz, R., Dvorak, D., Trandem, S., Berseth, B., and Lendway, M., "Utilizing UAV Payload Design by Undergraduate Researchers for Educational and Research Development," Proceedings of the 2007 ASME International Mechanical Engineering Congress and Exposition, IMECE2007-43620, November 2007.

3Semke, W., Ranganathan, J. and Buisker, M., "Active Gimbal Control for Surveillance using Small Unmanned Aircraft Systems," Proceedings of the International Modal Analysis Conference (IMAC) XXVI: A Conference and Exposition on Structural Dynamics, 2008.

4Dvorak, D., Czarnomski, M., Lendway, M., Martel, F., Semke, W.,Schultz, R., and Neubert, J., “Using Small UAVs to Capture Multispectral Imagery for use in Precision Agriculture.” Proceedings of the Association for Unmanned Vehicle Systems International (AUVSI) Unmanned Systems North America 2008, San Diego, CA, June 10-12, 2008.

5Semke, W., Stuckel, K., Anderson, K., Spitsberg, R., Kubat, B., Mkrtchyan, A., and Schultz, R., "Dynamic Flight Characteristic Data Capture for Small Unmanned Aircraft," Proceedings of the International Modal Analysis Conference (IMAC) XXVII: A Conference and Exposition on Structural Dynamics, 2009.

6Lendway, M., Berseth, B., Trandem, S., Schultz, R., and Semke, W., “Integration and Flight of a University-Designed UAV Payload in an Industry-Designed Airframe,” Proceeding of the Association Unmanned Vehicle Systems International, 2007.

7Marshall, D., Trapnell, B., Mendez, J., Berseth, B., Schultz, R., and Semke, W., “Regulatory and Technology Survey of Sense and Avoid for UAS,” Proceedings of AIAA Infotech@Aerospace, 2007.

8Lendway, M., Berseth, B., Martel, F., Trandem, S., and Anderson, K., (Schultz, R. and Semke, W., Faculty Advisors), “A University-Designed Thermal-Optical Imaging Payload for Demonstration in a Small Experimental UAS,” AIAA Infotech@Aerospace, 2007, (Selected as a finalist in the AIAA Intelligent Systems Student Paper Competition).

9“Creating Piccolo Aircraft Models with AVL” by Cloud Cap Technology.