Research On Path Planning Of Outdoor Mobile Robot Based On Multi Information Fusion

The cost map is constructed first in the path planning of mobile robot,and then the trajectory planning is carried out on this basis.Different mapping algorithms such as gmapping,cartographer,Hector and karto will show different effects in the process of path planning,among which,the cost map path planning of the cost map can be constructed by cartographer algorithm,But the trajectory of the starting point to the end point planning takes a long time;In the process of constructing and avoiding obstacles,the robot will cause large distortion error and even failure due to the over dependence on a single sensor;When a* and artificial potential field method are used in the cost map,the robot nodes will be loaded heavily,which will lead to false death and fall into local minimum points.In view of the above problems,this paper will be implemented in detail:(1)In view of the problem of poor path planning trajectory in different algorithm construction cost diagram,the two drive differential motion control of the lower robot locomotive body will be used to solve the problem.That is,PID control and BP neural network adaptive speed regulation are used to obtain mileage information.Then,the irregular distance information is synthesized into straight line by combining the linear fitting of odometer,Thus,the more accurate local coordinate position of robot is obtained;On this basis,GPS sensor is used to obtain the robot position in global coordinate system,so that the dynamic position and attitude information of robot can be accurately obtained.Finally,the optimal algorithm of constructing cost graph is verified by experiments in complex area scene of 5m x 80 m.(2)there is a big error in the cost estimation of the single sensor in the path planning of the laser radar.The simultaneous interpreting of the pose and pose of the sensors using IMU,ultrasonic,odometer and monocular camera and other sensors is based on the principle of priority of accuracy and the principle of different sensor fusion in different environments.Finally,the optimal position and attitude information of robot can be output,which can effectively avoid the problem of offset error caused by the low detection accuracy of single sensor.(3)In view of the problem of robot failure caused by using a* and improved artificial potential field method in cost map,dynamic update of node load and virtual target point method will be used to make robot walk out of pseudo dead and local minimum points respectively,Finally,based on this,the paper proposes a multi information fusion based algorithm to enhance the priority control of the lower locomotive body,and the upper computer Dijkstra algorithm to carry out path planning.The experimental verification is carried out through the experiment in the complex area scene 2 of the curved and right angle sharp turn road scene 1,the complex area scene 2 with 5m x 80 m,the scene with the long-distance path preferably fixed 8m x 120 m and the additional dynamic obstacle scenario 4,The improved algorithm is obtained from the starting point to the end of the traversal process.The path planning trajectory of the cost graph is shortened by 1.2% compared with that of karto algorithm;Compared with the cost map of cartographer algorithm,the cruise time of path planning is reduced by 5.7%,which meets the requirement of shortest smoothness of the path in the fixed complex area scene.