LEE- Savunma Teknolojileri Lisansüstü Programı
Bu topluluk için Kalıcı Uri
Gözat
Sustainable Development Goal "none" ile LEE- Savunma Teknolojileri Lisansüstü Programı'a göz atma
Sayfa başına sonuç
Sıralama Seçenekleri
-
ÖgeAI-based visual odometry implementation on an embedded system(Graduate School, 2023-06-12) Büyüksolak, Oğuzhan ; Güneş, Ece Olcay ; 514201027 ; Defence TechnologiesNavigation technology has always been a critical sub-field for defence technologies. The roots of navigation technology extend to early as 3000 BC year. The early examples of navigation technology were the observation of stars, bird following, etc. In modern days, Global Navigation Satellite Systems(GNSS) and Inertial Navigation System(INS) are widely used in defence technologies. The GNSS and INS systems are included nearly all the modern military platforms and smart munitions. However, they have also practical limitations. Therefore, a companion or alternative to these systems may provide flexibility in the navigational technology for military platforms. Visual odometry is an emerging technique for navigation technologies. The visual odometry technique is a dead reckoning technique, and it is the process of relative pose change estimation from camera images. Processing camera images for visual odometry is a computationally heavy operation. As the military platforms vary between a large scale of different sizes, their power and computational requirements also vary. Also, for expendable platforms like smart munitions, the cost is another important requirement. Embedded visual odometry(VO) implementation may provide a low-power, low cost and small-size alternative or companion positioning system to GNSS and INS. Hence the embedded systems are memory scarce, in this work, a new low-memory footprint neural network-based visual odometry method that is implementable on embedded systems is introduced and evaluated. In this work, firstly, the existing literature for geometry-based and deep learning-based methods was examined. Due to robustness and energy efficiency advantages, it was decided to realize a deep learning-based method, which can be further classified as supervised and unsupervised methods. The supervised learning methods generally require the involvement of other sensors than images and recurrent neural networks. These requirements come with additional computational and power consumption. As a very low-power and real-time system was targeted for this work, an unsupervised learning approach was selected as a training framework. With the ideas from the lightweight convolutional neural networks literature, a neural network namely TinyVO was designed. As monocular techniques provide more robustness and cost advantage than binocular ones, it was decided to use a monocular visual odometry technique in this work. The TinyVO network was trained with a well-known scale consistent unsupervised learning framework. After the training, TinyVO's performance on the KITTI dataset was evaluated. To deploy the neural network, MAX78002 artificial intelligence microcontroller has been chosen as the embedded platform. As MAX78002 supports 8 bits weights, the TinyVO was quantized to 8 bits using the provided MAX78002 software toolset. Also, the toolset supports the known answer test functionality. The known answer test was realized by using a sample from the KITTI dataset. Then, the energy and time consumption per inference values was measured. The resulting neural network, TinyVO enables a monocular visual odometry solution with a low-memory footprint, low power, and reasonable accuracy. To the best of found knowledge, this is the first study that provides a microcontroller-based visual odometry solution.
-
ÖgeLaser induced graphene based flexible gas sensor for wearable electronics(Lisansüstü Eğitim Enstitüsü, 2020-07-21) Büyükturgay, Mehmet Mert ; Solak, Nuri ; 514181008 ; Defence Technologies ; Savunma TeknolojileriFrom the past to the present, the most important thing to consider when developing defense systems is the production of original technology. The industrial power of a country is dependent on its technology. In this sense, it is important for countries to improve their original technologies. Technology is also important in the means of the privacy, reliability, and foreign dependency of defense systems. Production and development of high technology defense systems open a country's way to the domination of the world market, hence increasing the living standards of its citizens. This situation causes competitiveness in the field and in order to establish military domination, it is essential for countries to be able to produce and improve high-technology defense systems. Electronic technologies are also important in the defense systems industry. Since electronic technologies are widely used, engineers usually spend more shift hours designing and developing these products. There are lots of parameters in order to make these products ideal to be used in the defense systems industry. It is important and necessary for defense systems to be smaller in size, lightweight, with less effort and power required, practical, and faster in data transmitting. However, it may not always be easy to meet these necessities. In designing a product, just meeting one of these necessities may require a million dollars. The widespread use of electronic technologies results in the formation of new systems that are smaller, lightweight, and microchip controlled. In a changing and developing world, software updates of defense systems are also very important to function effectively. At the same time, the textile industry has recently begun to shift from traditional textile products to intelligent new textiles that process information to meet the demands and requirements of the defense industry. This situation resulted in engineering fields such as material science, electronics, and chemistry leading the textile industry and in the formation of a new multidisciplinary working area. Sensors compose almost the entire range of the smart textile field and unlike traditional textiles, it is a field open to new technological developments. Smart textiles are usually designed and developed for military technologies and the health sector. Wearable technologies complete duties such as obtaining and processing important data from the environment and simultaneously sending and visualizing these data to different sources. Smart clothing developed for area scanning and search and rescue teams may provide life safety for both search and rescue and military staff. They may also make any intervention more effective. Furthermore, information such as hazardous toxic gasses, amount of oxygen, and temperature of the potential risk zone and accident area may be obtained through this technology. This information is sent to different sources simultaneously and necessary measures are taken immediately. With the development of this system, risk analysis may be carried out in more detail, based on the information obtained from risky regions. The smart sensors may also be used in the exploration of the risky area; hence making the planning and application of the required intervention safer. Gas sensors play a major role in war fields and accident areas. Nowadays many countries are producing various chemical and biological weapons, so it is a necessity for every country to develop its own defense systems and take necessary precautions. New smart gas sensors are being developed as protection systems in case of the usage of these weapons. Gas sensors are not only used in the field of military defense systems; they are also used in companies where chemicals are frequently used and in highly populated regions in order to protect the health of workers and individuals by distinguishing hazardous and non-hazardous gasses in the air. Today, although many types and functions of commercial products are produced, problems in detecting hazardous gasses have not been solved completely. The problems that usually vary according to the type of sensor and need to be solved are as follows; the size of the sensor may not be small enough, low sensitivity and selectivity, long response time, not being suitable for long-term use, easy abrasion, sensitivity to movement, measurement errors due to environmental factors, high energy consumption, and difficulties in production and high costs. Briefly, more advanced and smart devices are needed, in order to detect what type of gasses are in the air. In this study, different sensor technologies and systems were studied to detect toxic or hazardous gasses. A major part of this study consists of research that accounts for precision, selectivity, response intervals, flexibility, resistance, size, and portability of the sensors. Even in their applications in daily life, sensors should be able to obtain data while they are in motion and transmit this data to distant sources. Therefore, in order to prevent the limitation of motion, the sensors should be portable and flexible enough to wrap the body easily. Wearable sensors that are designed compatible with clothing are more effective and useful than handheld sensors. The aim of this study is to determine whether the environment is dangerous or not in terms of gas type and to ensure that this information is transmitted to the source. Since this study is designed in the means of wearable technology, the disadvantage of carrying any device is eliminated. In addition, with the help of the sensor and resistance meter on the clothing, instant toxic gas contact can be detected. Therefore, staff on duty are able to obtain data from the environment simultaneously and can act more precautious to the current situation. First, a 3D graphene sensor was produced on the polyimide PI film using laser direct writing to create direct ammonia gas sensors. At this stage, first, the CO2 laser to be used has been optimized, the focus of the laser has been determined and the optimum parameters that will affect the final product have been determined. Then, carbonization and graphitization processes were applied according to these parameters. However, the necessary software was previously installed on the CO2 laser to implement this phase. With the help of the "Inkspace" application, sensors are designed and drawn. Then, with the help of the "K40 Whisperer" driver, the previously drawn sensors were transferred to the memory of the laser. The laser system was controlled with this driver, the predetermined parameters were entered into the system and the laser engraving process was started. As a result of this process, the desired laser-induced graphene products were obtained. The bending test was applied to the graphene sensors obtained. The products that were successful in this test were tested again for resistance with the help of a digital multimeter. The samples obtained after these tests were used in the specially designed NH3 gas sensor system. After the LIG sensor specimens were obtained, the chamber was ready to be installed. NH3 gas sensor mainly consists of a bottle of ammonia gas (25%), a micro syringe, control switches, a jar for the gas mixing system, a jar for the chamber, 1 air fan, 2 pneumatic pipes (1 for gas inlet and 1 for gas outlet), a digital multi-meter and pre-produced graphene sensors. This test was done at room temperature and room atmosphere. First, a micro syringe was filled with a certain volume of NH3(25%); then it was injected into a gas mixing system. A certain concentration of ammonia gas(100ppm-1000ppm) was obtained by mixing with ambient air in the mixing system. After that, the first switch was turned on and the mixing gas vapor was transferred to the test chamber near the LIG sensor. The LIG specimen was placed inside the chamber. In this system, one of the control switches was used to transfer a certain concentration of ammonia gas to the chamber and the other switch was used to release the gas from the chamber. Thus, a certain concentration of ammonia gas was circulated in the chamber homogenously with the help of the air fan. The digital multimeter that was used in this chamber is able to show the resistance of the sensor when it was exposed to ammonia gas and the atmosphere. As a result of the changing ambient atmosphere, the resistance meter connected to the graphene sensor showed a resistance change due to NH3 gas contact. Therefore, ammonia gas sensing was carried out. As a result, a flexible graphene-based gas sensor was produced, which detects whether there is any ammonia gas as a warfare gas simulant in the atmosphere. The produced graphene-based gas sensors can operate at room temperature with high selectivity and low power consumption. Thus, this study can be an important candidate for today and future applications of wearable detectors in both military and many other fields.