Research On Key Technologies Of Autonomous Navigation Intelligent Patrol Car

Autonomous navigation in the regional environment is a research hotspot of current intelligent robots.The patrol car based on autonomous navigation can replace the staff to perform daily patrol tasks,and carry out emergency rescue and related service work in dangerous places,it has received extensive attention from relevant industries at home and abroad.The autonomous navigation control method involves key technologies such as environmental perception,behavioral decision-making and motion control,and there is a problem of insufficient prior information in the regional environment,which must be solved by combining positioning,motion modeling and planning control methods.In view of the problems of positioning and autonomous navigation capabilities of patrol smart devices,this thesis studies the key technologies of autonomous navigation patrol cars based on high-performance embedded architecture.The content mainly includes regional environmental classification and path planning methods.Through the combination of lidar and image edge detection methods,the functions of distinguishing the path and obstacles,establishing regional maps and selecting paths are completed.In terms of regional environmental classification,the regional positioning of the car is first completed in the actual environment lacking a priori information by Monte Carlo filtering.Secondly,aiming at the problem of light and darkness in the environment,the existence of error in the boundary of obstacles and the low credibility of single sensor detection,it is proposed to improve the regional classification credibility by combining image and lidar.On the one hand,the image edge detection is performed by the Canny operator based on the road reference width and parallel line detection,and the edge of the road surface and the obstacle is extracted;on the other hand,the contour information of the environment is collected by the laser radar high-frequency information,and the image is extracted.The edge information is compared,and the adaptive weighted averaging method is used to achieve data fusion to improve the credibility of the edge of the path.For the regional map establishment,a particle filter-based SLAM method is used to establish a more accurate patrol environment map.In terms of the path planning method,combined with the overall and local characteristics of the environment,the combination of global and local path planning is used for analysis.In the aspect of global path planning,the principle andimplementation effect of the classic Dijkstra and A* algorithms are analyzed.Aiming at the contradiction between search scope and security,this paper adopts A* algorithm based on extended neighborhood search range and improved direction guidance and exponential decay heuristic function,which improves the effect of planning path in ensuring traffic safety.For the local path planning,the sliding window(DWA)method is adopted to increase the sampling speed and predict the trajectory in the dynamic environment.At the same time,the UWB label is set as the reference node of the complex zigzag inflection point and the like,and the patrol car travel path is performed.Correct it in time to achieve better planning results.Through the above-mentioned regional roadblock resolution and map establishment,path planning and other key methods,this paper can be found that it can effectively pass the complex environment and avoid obstacles,and better complete the autonomous navigation task in the region.