Design and Development of the UTSA Unmanned Aerial System ACE 1 For use in the 2010 AUVSI Student UAS Competition Ilhan Yilmaz Department of Mechanical Engineering (Team Lead) Christopher Weldon Department of Electrical and Computer Engineering Nihat Altiparmak BS.CE Department of Computer Science
1 Table of Contents 1 The UTSA UAV Team... 2 2 Team Composition... 2 3 Objectives... 3 4 Airframe Development... 3 4.1 Test UAV... 4 4.2 Competition Airframe... 5 5 Systems Development... 9 5.1 Auto-Pilot... 10 5.2 Camera and Image Recognition... 14 5.2.1 Camera... 14 5.2.2 Ethernet Bridge... 14 5.2.3 Software... 15 6 Safety... 16 7 Final Statement... 17
The UTSA UAV Team 2 1 The UTSA UAV Team This marks the first year UTSA has entered the Student UAS competition. To compete we have created a fully autonomous UAV capable of carrying a large surveillance payload. More specifically, the UAV boasts a 10 foot wingspan, and almost the same in fuselage length. The plane is under primary control by the MicroPilot 2128 LRC auto-pilot system. Using this, the aircraft is capable of fully autonomous takeoff, landing, flight, and complies with all safety requirements in accordance with the competition rules. Aerial surveillance is achieved through the use of a fully positional 20x optical zoom video camera. 2 Team Composition The UTSA UAV team is composed of two undergraduates in Mechanical and Electrical Engineering, and a Graduate Computer Science student.. The team operates out of the Autonomous Control Engineering Lab of the Department of Electrical and Computer Engineering at UTSA. Project responsibilities were coordinated respectively, the first partition being the mechanical design and development of the airframe. Secondly, the electrical split involved implementation of the auto-pilot, control software, power management, and overall communication systems. The computer science additive of image processing and recognition completed the requirements of this project.
Objectives 3 3 Objectives In accordance with the competition rules, we set out to accomplish these objectives. Autonomously detect and collect data on targets (location, time, orientation, shape, shape color, alphanumeric symbol, and symbol color). Then, present that data in an actionable format in accordance to the format specified in the rules. The system shall be able to display a planned waypoint course and also show course boundaries and no-fly zones. The system should also be capable of presenting real time information on the aircrafts position, speed, and altitude in relation to the course and to ensure the UAV is avoiding the no-fly zones. The UAV shall be capable of autonomous takeoff and landing. The UAV shall be capable of conducting aerial surveillance on a predetermined course, and shall be capable of correcting that course and search parameters during flight. The camera shall be capable of three images per second in order to detect the required targets. The system shall comply with all safety requirements set forth in the rules. 4 Airframe Development Two airframes were used in the development of the UAS. A smaller stock airframe was ordered with the autopilot as a test vehicle. This airframe was capable of many of the functions we needed for competition, mainly complete autonomous control. It is also capable of mounting a camera for use in the surveillance objectives. However, the test craft was not capable of carrying the Toshiba IK WB21A camera, nor did it have the payload capacity to hold the Microhard modem that would be used to transmit the video feed. Therefore, we developed a competition airframe with a significantly larger wingspan and payload capacity. The larger airframe also meant we could decrease our minimum speed if necessary to conduct accurate ground surveillance. The extra space, allows room for expandability in future developments as necessary.
Airframe Development 4 4.1 Test UAV During the development stage of the Competition airframe briefly described above, a smaller test airframe was used. More specifically, we used the Alpha 60 trainer airframe that arrived stock with the order of the MP2128 LRC auto-pilot system. This was relatively simple to initiate, requiring very little in the terms of build time, and implementation of the sensors and other technical equipment. The point of this trainer UAV was to test the auto pilot system and calibrate it to our needs for competition. The other objective was to evaluate its performance and decide what necessary settings needed to be altered when transferring the auto-pilot system to the significantly larger competition airframe. This airframe also served its purpose as a test suite for the Horizon Software used to control the auto-pilot functions. The trainer was not equipped with a surveillance package, and therefor was only suitable to test flight characteristics and evaluate the auto-pilot s performance. Figure 1 Alpha 60 Trainer
Airframe Development 5 4.2 Competition Airframe The competition airframe in use is custom built for the 2010 Student UAS competition. The airframe bares some Military UAV resemblance. The main feature of this design is a rear mounted engine. This enables the airframe to have reduced vibrations in the front where the sensory, surveillance, and auto-pilot equipment is located. Having the engine mounted in the rear also makes the plane more aerodynamic, in the sense that any drag that would have been created by the engine is no longer an issue due to the fact that the fuselage now covers the engine s non-aerodynamic surfaces. Due to the fact that the engine came forward configured, we run it in reverse, and are able to do this by adjusting the timing of ignition. Due to the rear engine configuration, the tail is built on two carbon fiber rods, mounted below the wings. The tail will have a dual rudder configuration, and a single elevator. Overall, large control surfaces were used in this airframe to help with stability and achieve higher degrees of control at low speeds. The aircraft also has a thrust to weight ratio higher than ½. This ensures optimum power to support our payload and low speed capabilities. Figure 2 Competition Airframe Construction The larger wingspan allows for greater stability in low speed, sustained flight. The larger fuselage enables us to carry up to 10lbs of cargo. Specifically, the fuselage is 8x8x36 inches. With the exception of some concavity in the front of the airframe, this allows us almost 2,304 cubic inches of payload space. Cargo includes the camera, batteries, auto-pilot, wireless link, and a 150cc fuel tank. We decided
Airframe Development 6 to use batteries to power all of the electronics. In the original design, a generator was to be used to provide a power plant for the electronics. Due to the fact that the objectives of the mission were to be completed in 20-40 minutes, using batteries instead, proved it to be lighter. All electronics will be wrapped in foam and securely mounted to the fuselage in order to reduce vibration and improve safety characteristics. Batteries are of course brightly colored in the event of an aerial failure that results in the loss of the aircraft. Figure 3 Fuselage Interior Airframe was built around materials such as balsa and plywood, and is reinforced with carbon fiber and fiberglass, which adds to the strength and rigidity of the airframe, and allows for the appropriate amount of flexibility. The fuselage is constructed from a balsa and plywood mix, and coated in fiberglass. The main structural support and connection from the fuselage to each wing is one 1.125 carbon fiber tube that extends across the entire airframe, halfway into each wing. There are also carbon fiber and fiberglass reinforcements used throughout the construction of the aircraft where it was necessary to add to the rigidity of the airframe. The overall material construction allows for a relatively light airframe considering its size. The target max weight is 25lbs. With 10lbs allotted for the payload, this leaves 15lbs allotment for the airframe weight. The strength that is designed into the airframe, not only in material selection, but in construction techniques is more than necessary to accomplish its surveillance function. This allows us to recover from unexpected maneuvering, in the event of environmental conditions becoming less than optimal. Other design choices include a wide base on our landing gear. The displacement between the base landing gear is 31. The landing gear
Airframe Development 7 supports are manufactured out of carbon fiber and extend from the rear fuselage of the aircraft. The front gear is designed to be steerable and withstand a significant amount of shock. The props used are manufactured from carbon fiber and wood, and have diameters of 21. All sculpted parts of the airframe (nose, etc.) are constructed from lightweight hard foam, and coated in fiberglass for rigidity. Figure 4 Wing construction and support Figure 5 Wing profile and final shape
Airframe Development 8 Table 1 Specifications Fuselage Length 87 in Width 8 in Height 8 in Wingspan 116 in Chord 13.5 in Area 10.5 ft 2 Ailerons 4x24 in Engine DA-50cc Fuel 87 octane gas Oil Mix 1:40 oil:gas ratio
Systems Development 9 5 Systems Development In terms of communication systems, the UAV has three communication channels, two 2.4GHz broadband data links for controls and surveillance, and one 72MHz radio link for emergency override and control. The first 2.4GHz data link is used for primary control of the aircrafts auto-pilot system and as a sensory data relay. This channel is used exclusively by the auto-pilot to insure a dedicated connection and all necessary bandwidth. Through the autopilot communications channel we can manually take control of the aircraft in the event of an auto-pilot failure and only if the data link has not been compromised. The second 2.4GHz data link is dedicated to the camera surveillance system. This ensures we have optimum bandwidth to transmit live imagery at high speeds between the aircraft and base station. Figure 6 Block Diagram of UAV Systems In the event of a total failure of the auto-pilot communications link, we can take manual control of the aircraft by means of a 72 MHz radio
Systems Development 10 link and an onboard MUX (Multiplexer) that at the flip of a switch, can take complete control of all essential controls functions of the UAV. Overall, this gives us a system with multiple levels of redundancy. 5.1 Auto-Pilot The autopilot in use by the UAS is the MicroPilot 2128 LRC. This autopilot uses GPS dependent fly-by-way-point technology in order to navigate. The 2128 LRC has a full avionics and sensory package that includes autonomous surface control of all aileron, flap, elevator, and rudder surfaces, steerable front landing gear, as well as throttle control. The autopilot is capable of making decisions based upon changing conditions that would affect airspeed, position, and altitude. Autonomous take-off and landing is achieved via the addition of an AGL ultrasonic sensor mounted on the underside of the airframe. This sensor is accurate to the inch with a 14' maximum range. This precision allows for accurate landings, necessary due to the airframes size. Figure 7 (From left to right) AGL Board, AGL Sensor, GPS Antennae Communication is achieved via a 2.4GHz radio link via remote and ground modems. Only navigation and control communications are allowed across this particular set of modems to insure a dedicated and uninterrupted communications channel to maintain reliable control of the aircraft. This channel handles all control commands, sensory data, and telemetry data received from the GPS and avionics packages, as well as data received from the AGL unit. The video feed is handled by its own set of modems and communications channels.
Systems Development 11 Figure 8 Remote Autopilot/Modem and Ground Modem Data is collected at a frequency of 5Hz (5 records per second). This data is communicated through the Horizon control software for the 2128 LRC and recorded in a designated log folder on the base station computer. This data can then be accessed and later analyzed for use in time stamping and recording intelligence on the collected images taken from the camera. Data may be configured by the Horizon software to record and transmit in any fashion, allowing us to keep customized logs with only the necessary information for a specific purpose (I.E. image recognition and recording processes), while also maintaining separate complete logs that are useful in debugging control methods and algorithms. Figure 9 Horizon Control Software
Systems Development 12 Way-point programming and control is done through the Horizon software as well. Way-points are placed on an imported map; scale compensation and geographic calibrations are done during the importing process. Way-point files can be generated pre-flight and edited on demand throughout the flight. Way-points can be altered, added, and deleted at any point, allowing for full control of the UAS during any stage of flight. There are variations in the fly commands that can be used to travel between way-points. Specifically, the "flyto" command can be used to fly between destinations, although, the route it used to get there is not necessarily important. Meaning, the UAS will reach its destination accurately, but say if conditions keep the plane from flying a straight course; the autopilot will not continuously correct its course. Alternatively, the "flyfrom" command may be used to treat the path between destinations as relevant, and will course correct to follow the designated path more accurately. Figure 10 fly-by-waypoint operations The Horizon software is capable of calibrating all servo communications and controls, allowing the autopilot to have a high responsiveness to changing conditions and sensory outputs. Various throttle percentages, control surface angles can all be calibrated for the various functions of flight, take-off, and landing.
Systems Development 13 Figure 11 Horizon Displaying Servo PID Settings The software is also capable of conducting flight simulations with weather conditions. The simulation treats the Horizon software as if it was in flight, and allows maps, way-point files, emergency protocols, as well as take-off and landing procedures to be tested before the flight actually takes place. Figure 12 Horizon Simulation Settings
Systems Development 14 5.2 Camera and Image Recognition 5.2.1 Camera We decided to use the Toshiba IK-WB21A (Figure 13) pan-tilt-zoom network camera in our target detection system for several reasons. The most important criteria we considered were transferring the images from the plane to the ground station easily and fast. In our beliefs, the best way to manage this is using a separate data link for the camera and sending the images through this link directly using the capabilities of a network camera. By this way, we do not need to use any intermediate software for the camera driver and any other processing unit on the plane. Toshiba IK-WB21A is a network camera which can be connected to a Local Area Network (LAN) with an Ethernet cable and can be reached from any computer connected to the same LAN using its IP address, or possible from any computer connected to Internet if the LAN the camera connected has an Internet access. In other words, we can easily manage the camera from the ground station using a wireless Ethernet bridge between camera and the plane. Secondly, we considered its pan-tilt-zoom and image quality capabilities. The camera can take 7.5 frames per second with the resolution of 1280 x 960 and has 22x optical zoom capability, which will be enough for our target detection system. Figure 13 Toshiba IK-WB21A 5.2.2 Ethernet Bridge We used a separate fast wireless data link for the target detection system by using Microhard Systems' VIP2400 wireless Ethernet Bridge (Figure 14). It comes with two devices, one is called the station, which stays on the plane (air device), and the other is called
Systems Development 15 access point which stays in the ground station (ground device). The Ethernet Bridge uses Orthogonal Frequency Division Multiplexing (OFDM) to provide very fast, robust, and long range data transfer at speeds of up to 54MbpsI. The bridge has firewall and encryption support with QoS (Quality of Service) assurances. Another advantage of the Ethernet Bridge is that multiple stations can be connected to one access point by allowing a system with multiple UAVs and one ground station, which improves the scalability of the system. For the setup of the imaging system, it is enough to connect the IP camera to the LAN port of the station with an Ethernet cable. Although it comes with a high gain antenna, it is possible to use higher gain antennas to improve the longer range data transfers. Figure 14 Microhard Systems VIP2400 wireless Ethernet Bridge 5.2.3 Software We implemented target detection software using Matlab and its image processing toolbox. Our camera takes 7-8 images per second and puts them in a shared user specified folder with their time stamp information written on their filenames. Target detection software gets the folder name as an input, processes all the images in that folder and waits for the new images to arrive till user stops the software. In order to detect the target, series of edge detection, threshold and color detection techniques are applied using mostly the predefined functions of Matlab image processing toolbox. After detecting the target on an image, the autopilot's log file is used to match the GPS and orientation information belonging to each specific target. Autopilot logs the information of time, GPS_E, GPS_N, pitch, roll, air speed, GPS speed, altitude and heading
Safety 16 information in a log file with the frequency of 5Hz. We know the time when each image is taken. By using this time, we select the closest time of log information and save this information into another file by tagging it with the related image file. Since we know the GPS and orientation info for each target, we can make the calculation necessary to find the exact GPS location and orientation of the target by using the pixel location of the target on the fly. 6 Safety Safety was a primary factor in the design of this aircraft. Careful consideration was taken to ensure that all of the competition rules regarding safety were followed. Batteries have been covered with bright electrical tape to ensure their visibility in the event of a crash. Components are securely mounted to ensure that they remain secure in the airframe. They are also wrapped in thick soft foam to reduce vibration that would cause them to malfunction and/or even cause system failure. In terms of controls, an autopilot communication failure after 30 seconds will cause the autopilot to return home. If the GPS were to fail for more than 30 seconds, the emergency override would be enabled by the safety pilot, and the plane can be brought down manually by a separate RC transmitter. In the case of a complete communication failure (including redundancy control and GPS), after a period of 3 minutes, the autopilot will enter into a pre-determined flight procedure (listed in the rules) and terminate itself. The autopilot is capable of staying on a predetermined waypoint course, and accepting command while in flight to ensure it stays inside the course boundaries and outside of the no-fly zones. Specifically, if there is not enough room for the plane to be affected by flight conditions and take a curved path, fromto commands can be utilized that will force the autopilot on a straight line path between destinations.
Final Statement 17 7 Final Statement With our unique airframe and redundant, safety oriented systems approach, we plan on being major contenders. Overall, we expect this competition to be the beginning of many great experiences.