| dc.description.abstract | Position navigation and timing (PNT) is the concept of determining where an object is on the Earth (position), the destination of the object (navigation), and when the object is in these positions (timing). In autonomous applications, these three attributes are crucial to determining the control inputs required to control and move the platform through an area. Traditionally, the position information is gathered using mainly a global positioning system (GPS) which can provide positioning sufficient for most PNT applications. However, GPS navigational solutions are limited by slower update rates, limited accuracy, and can be unreliable. GPS solutions update slower due to the signal having to travel a great distance from the satellite to the receiver. Additionally, the accuracy of the GPS solution relies on the environment of the receiver and the effects caused by additional reflections that introduce ambiguity into the positional solution. As result, the positional solution can become unstable or unreliable if the ambiguities are significant and greatly impact the accuracy of the positional solution. A common solution to addressing the shortcomings of the GPS solution is to introduce an additional sensor focused on measuring the physical state of the platform. The sensors popularly used are inertial measurement units (IMU) and can help provide faster positional accuracy as the transmission time is eliminated. Furthermore, the IMU is directly measuring physical forces that contribute to the position of the platform, therefore, the ambiguities caused by additional signal reflections are also eliminated. Although the introduction of the IMU helps mitigate some of the shortcomings of GPS, the sensors introduce a slightly different set of challenges. Since the IMUs directly measure the physical forces experienced by the platform, the position is estimated using these measurements. The estimates of position utilize the previously known position and estimate the changes to the position based on the accelerations measured by the IMUs. As the IMUs intrinsically have sensor noise and errors in their measurements, the noise errors
directly impact the accuracy of the position estimated. These inaccuracies are further compounded as the erroneous position estimate is now used as the basis for future position calculations. Inertial navigation systems (INS) have been developed to pair the IMUs with the GPS to overcome the challenges brought by each sensor independently. The data provided from each sensor is processed using a technique known as data fusion where the statistical likelihood of each positional solution is evaluated and used to estimate the most likely position solution given the observations from each sensor. Data fusion allows for the navigation solution to provide a positional solution at the sampling rate of the fastest sensor while also limiting the compounding errors intrinsic to using IMUs. Synthetic aperture radar (SAR) is an application that utilizes a moving radar to synthetically generate a larger aperture to create images of a target scene. The larger aperture allows for a finer spatial resolution resulting in higher quality SAR images. For synthetic aperture radar applications, the PNT solution is fundamental to producing a quality image as the range to a target is only reported by the radar. To form an image, the range to each target must be aligned over the coherent processing interval (CPI). In doing so, the energy reflected from the target as the radar is moving can be combined coherently and resolved to a pixel in the image product. In practice, the position of the radar is measured using a navigational solution utilizing a GPS and IMU. Inaccuracies in these solutions directly contribute to the image quality in a SAR system because the measured range from the radar will not agree with the calculated range to the location represented by the pixel. As a result, the final image becomes unfocused and the target will be blurred across multiple pixels.
For INS systems, increasing the accuracy of the final position estimate is dependent on the accuracy of the sensors in the system. An easy way to increase the accuracy of the INS solution is to upgrade to a higher grade IMU. As a result,
the errors compounded by the IMU estimations are minimized because the intrinsic noise perturbations are smaller. The trade-off is the IMU sensors increase in cost, size, weight, and power (C-SWAP) as the quality of the sensor increases. The
increase in C-SWAP is a challenge of utilizing higher grade IMUs in INS navigational solutions for SAR applications. This problem is amplified when developing miniaturized SAR systems. In this dissertation, a method of leveraging the benefits of data fusion to combine multiple IMUs to produce higher accuracy INS solutions is presented. Specifically, the C-SWAP can be reduced when utilizing lower-quality IMUs. The use of lower quality IMUs presents an additional challenge of providing positional solutions at the rates required for SAR. A method of interpolating the position provided by the fusion algorithm while maintaining positional accuracy is also presented in this dissertation.
The methods presented in this dissertation are successful in providing accurate positional solutions from lower C-SWAP INS. The presented methods are verified in simulations of motion paths and the results of the fusion algorithms are evaluated for accuracy. The presented methods are instrumented in both ground and flight tests and the results are compared to a 3rd party accurate position solution for an accuracy metric. Lastly, the algorithms are implemented in a miniaturized SAR system and both ground and airborne SAR tests are conducted to evaluate the effectiveness of the algorithms. In general, the designed algorithms are capable of producing positional accuracy at the rate required to focus SAR images in a miniaturized SAR system. | en_US |
