Visual Inertial Navigation Method for Unmanned Mining Truck in Open-pit Coal Mine under GNSS and SINS Fusion
When unmanned mining trucks work in open-pit coal mine,they rely on a single strapdown inertial navigation system(SINS)to achieve vehicle inertial navigation.Over the long term,information accumulates during the navigation process,resulting in significant absolute position errors in the navigation trajectory.Therefore,a visual inertial navigation method for unmanned mining truck in open-pit coal mine was proposed under the fusion of global navigation satellite system(GNSS)and SINS.Obtained dynamic visual images of open-pit coal mine scenes,established a vehicle driving dynamic environment map based on simultaneous localization and map building(SLAM)algorithm,and used a judgment method based on 3D distance error to screen dynamic feature points and annotate dynamic areas within the map to locate the real-time position of unmanned mining truck.Integrated the GNSS and SINS to construct a combined visual inertial navigation model,and introduced the Kalman filter equation to optimize the model parameters.Finally,with the assistance of recurrent neural networks,the navigation results were corrected to generate high-precision navigation results for unmanned mining truck.The experimental results show that the absolute position error of the navigation trajectory formed by this method does not exceed 3 m,which proves its superior navigation performance.
NETL
NSTL
万方数据
