Study On CA-BINN Dynamic Path Planning And Visual Guidance Analysis For Gravel Soil Core Paving

The paving path planning of gravel soil core is very important for the quality and efficiency of paving process of core rockfill dam.However,most of the existing paving operation paths adopt static planning,which fails to consider the dynamic changes of obstacles in complex operation environment such as bulldozers and dump trucks,and the path planning results are separated from the paving quality analysis and the actual construction scene,which is lack of intuition and real-time.Therefore,this paper proposes the CA-BINN dynamic path planning and visual guidance analysis method for gravel soil core paving.The main research results are as follows:(1)Proposed a real-time reconstruction method for complex dynamic operation environment based on the theory of cellular automataIn order to accurately reflect the complex dynamic operation environment and provide environmental input information for CA-BINN path planning model,considering the characteristics that cells in CA can dynamically store rich information,based on CA modeling theory and combined with environmental parameter perception system,the real-time perceived complex operation environment data is abstracted into dynamic,obstacles paving thickness and other cell states to reconstruct the dynamic construction environment in real time.(2)Proposed a dynamic path planning method for bulldozer paving operation based on CA-BINN modelAiming at the problem that the existing paving operation path planning methods fail to consider the dynamic changes of obstacles in complex operation environment such as bulldozers and dump trucks,the cell state with information such as dynamic obstacles and paving thickness is used as the external input of biological inspired neural network algorithm,The neural activity value of BINN algorithm under the CA framework is reconstructed to calculate the diversion equation to realize the dynamic path planning of multi pile overall paving operation.The single pile paving operation adopts the three knife method CA rules proposed by the on-site three knife method(three times pushing and moving pile)construction process to realize the dynamic path planning.(3)Proposed a 3D visualization trajectory and quality control method based on augmented reality technologyAiming at the problem that the existing bulldozer paving operation planning path results are separated from the paving quality analysis and the actual construction scene,which is lack of intuition and real-time,a paving three-dimensional visual trajectory and quality control method based on augmented reality technology is proposed,which couples the planning path with the actual scene for automatic drawing and guidance of the trajectory,Combined with the construction site,the real-time analysis of paving quality is carried out,so as to realize the dual control of track and quality.Taking a core rockfill dam in Southwest China as an example,an augmented reality visual trajectory and quality control system is developed,and the simulation and field application based on the system are carried out.The results show that the proposed method can not only show high safety and high adaptability in complex dynamic operation environment,but also reduce the path length,number of turns and invalid path ratio decreased by 1.9%,42.9% and 48% respectively,and the paving flatness increased by 7%;Compared with manual operation,the paving flatness is increased by28% and the invalid path ratio is reduced by 47%,which effectively improves the paving quality and paving efficiency.