Research On Visual Odometry/IMU Assisted GPS Fusion Location Algorithm

Different sensors are used for navigation and positioning,while each has its own advantages and disadvantages.According to different application scenarios,different sensors are often integrated to obtain optimal navigation and positioning performance.GPS signals are prone to interruption and signal quality is poor in urban,bridge and other occlusion environments,which results in low accuracy and reliability of positioning results.So this thesis will mainly study the visual odometry/IMU-assisted GPS integrated positioning algorithm to improve the navigation positioning accuracy and reliability in outdoor restricted environment and provide technical support for the upcoming unmanned era.The main works were as follows:1.The key technologies of visual odometry(VO)and inertial navigation unit(IMU)were discussed separately.Aiming at the large amount and mismatching of feature point matching in visual odometry,the preprocessing algorithm was proposed that based on Euclidean distance threshold setting in the field of integrated image processing.It has been proved by experiments that the correct matching point quality of the picture is improved.2.Discussing the theoretical basis of VO/IMU integrated navigation algorithm and establish a VO/IMU tight coupling model.For the multi-sensor time synchronization problem in the experiment,this contribution used to estimate the time offset between hardware devices by software algorithm.3.A GPS/VO/IMU integrated algorithm based on robust adaptive Kalman filter was proposed.The GPS data was collected by the handheld receiving device,and the GPS/VO/IMU information was integrated to solve the carrier position.In the occlusion environment,compared to that of GPS-only solutions,the RMS values derived from GPS/VO/IMU solutions in E,N and U were improved from 9.7653 m to 4.48 m by 54%,from 31.8248 m to 7.55 m by 76%,and from 20.8644 m to 5.62 m by 73%,respectively.