Browsing by Author "Noureldin, Aboelmagd MA"
Now showing 1 - 3 of 3
Results Per Page
Sort Options
Item Open Access Deep Neural Network Aiding Visual Odometry for Land Vehicles Navigation(2020-12-09) Salib, Abanob M. A.; El-Sheimy, Naser M.; El-Sheimy, Naser M; Noureldin, Aboelmagd MA; Kattan, LinaSelf-driving cars consider vision sensor (monocular/stereo camera) as the primary sensor for driving vehicles and providing rich visual information which can be utilized for obstacle avoidance and scene understanding. This thesis introduces an improved visual odometry algorithm for vehicle navigation by including deep neural network such as YOLOv3 through masking the moving objects within each frame, and excluding these objects in order to aid the RANSAC (RANdom SAmple Consensus) to raise the inliers ratio. Pedestrians and moving vehicles can add outliers and reduce RANSAC performance. In some cases, the RANSAC algorithm can fail, such as when any dataset has a significant number of contaminated points or is non-realistic, such as within the dynamic environment. By integrating a machine learning module, RANSAC was able to rely more on static features than on dynamic features, resulting in lower RANSAC computing cost. Different datasets were used to check the proposed algorithm’s efficiency. The results are promising because they reflect a rise in the elapsed time reduction for primary matching and RANSAC and incrementation in inliers proportion. Through implementing the suggested approach, the final navigation solution for two different datasets presents a significant improvement over the typical visual odometry technique.Item Open Access Enhancing Land Vehicle Navigation in Challenging Environments Using Consumer Level Devices(2020-11-20) Moussa, Mohamed; El-Sheimy, Naser; Noureldin, Aboelmagd MA; Moussa, Adel; Helaoui, Mohamed; Gao, Yang; El-Mowafy, AhmedRecently there has been a massive effort in developing navigation systems for the self-driving cars. GNSS/INS integration is the most common sensor fusion technique to estimate the land vehicles navigation states. However, this system is not perfect in all operating situations as GNSS signals may suffer from signal outages, and/or multipath in urban and foliage areas. In such cases, INS provides the navigation solution which is degraded after a very short period due to the large drifts of the INS. During GNSS signal outage, INS should be assisted with other aiding sensors to mitigate its large drift. These sensors may include magnetometers, odometers, cameras, Light Detection And Ranging (LIDAR), Radio Detection And Ranging (RADAR), etc. Maps aiding navigation is used in many previous researches to help low-cost INS in GNSS denied environments. Consumer Portable Devices (CPDs) are widely used all over the globe. CPDs contain many sensors that could particpate in the enhancement of the land vehicles. Unfortunately, there are some limitations associated with the previous mentioned aiding techniques related to their high price, high computation and processing cost, weather and surrounding environmental effects in addition to the map aiding method drawback that is based on the avialability and the update rate of the required maps. Therefore, autonomous land vehicles navigation using low-cost sensors integrated systems has attained a lot of research interest The main objective of this research is to develop various land vehicle navigation systems that work in GNSS challenging environment, based on low-cost, non conventional sensors, and CPDs to add a redundant land vehicle motion information, reduce the cost of the navigation systems and thus decrease the overall self-driving car cost and provide an accepatble navigation performance. Two low-cost non-conventional wheel odometry systems are proposed where the first system is based on multiple ultrasonic sensors while the other is developed using multiple low-cost gyroscopes. These systems are implemnted to aid the low-cost INS in GNSS signal outage to reduce its drift. The relative alignment between the CPD and vehicle frames is very vital process when using CPD in the land vehicle navigation. However, it requires other sensors to estimate the relation between the CPD and vehicle coordinaite systems and it should be in motion. A new static relative alignment method is developed based on the vehicle vibration signature pattern which does not require any additional sensors. Moreover, a steering wheel angle estimation method is developed using CPD self-contained accelerometers that is attached to the steering wheel where this information is used in aiding the low-cost INS in GNSS challenging environment. Furthermore, different DR and aiding systems are investegated based on the land vehicle information and the CPD sensors to assess the navigation performance for such systems. Moreover, multiple CPD navigation systems are investegated using federated fusion technique.Two non-conventional navigation systems are proposed where the first depends on the ECU sensors to provide a redundant land vehicle speed information to aid the low-cost INS in GNSS signal outage. The second system is based on mass flow sensors that benefits from the aerodynamics of the vehicle motion to provide the speed and the heading information in indoor environment.The performance of the proposed low-cost navigation techniques show promising navigational results and low complexity efforts.Item Open Access In-Receiver Analysis and Estimation of GNSS Biases(2021-08-27) Wang, Ye; Gao, Yang; Zhao, Lin; EI-Sheimy, Naser M; Noureldin, Aboelmagd MAGlobal navigation satellite system (GNSS) biases significantly degrade performance when using GNSS for precise positioning, ionospheric remote sensing, and time transfer. To date, research on GNSS biases has been primarily based on measurements output from receivers. This thesis investigates the concept and method of bias analysis and estimation within receivers with a focus on the analysis and estimation of receiver-induced biases, the correlator spacing influence on single-frequency positioning, fractional cycle bias estimation, and differential code bias estimation based on a correlator spacing flexible software receiver. The in-receiver bias analysis and estimation approach will open doors for bias calibration within receivers to output measurements with biases eliminated. The calibration of GNSS in-receiver front-end local oscillator, chip, ADC, and baseband correlator spacing-induced biases has been investigated and analyzed for the same frequency. The results show that the front-end-induced biases cannot be ignored and cannot be simply included in a certain error. The correlator spacing-induced biases have significant differences between satellites based on an analysis of both simulated signals and live signals. The between-spacing real-time differential positioning method (BSRTD) and the between-spacing real-time kinematic positioning method (BSRTK) have been developed to improve the performance of single-receiver single-frequency positioning after an analysis of inter-spacing bias influence on single point positioning (SPP). The results show that the BSRTD and BSRTK methods can significantly improve the stability of single-frequency positioning with a smaller standard deviation, and the BSRTK method can offer faster ambiguity resolution. A cascaded FCB estimation method, named the multi-spacing single-receiver estimated FCB (MSFCB) method, has been developed based on different correlator spacing pseudorange and carrier phase observations. The MSFCB method can be used in the ambiguity resolution process. A software receiver-based multi-spacing DCB estimation method has been proposed to improve DCB real-time performance. By comparing the MSDCBs to the IGS DCB products, the results show that the proposed correlator spacing flexible software receiver is able to estimate satellite DCBs with increased flexibility and cost-effectiveness compared to the current hardware receiver-based DCB estimation approach.