Research On Path Planning Of Mobile Robot

Robots have made great progress and development since their birth.Among them,the mobile robot that can move autonomously in the work space is developing rapidly,becoming a flourishing branch of robotics.Path planning is a classical problem in mobile robotics,it requires a non-collision path from the robot's starting position to the target position.This paper mainly studies the path planning of mobile robot under the static environment.Aiming at the problem of local path planning,a corner prediction method which combines the binary k-means algorithm with the convolutional neural network was proposed.This paper introduces the model in detail and designs experiments to prove that it has advantages in accuracy and speed of mobile robot corner prediction compared with the convolutional neural network model without clustering algorithm.For the global path planning,this paper improved the multi-step ant colony algorithm and proposed some innovative points:Firstly,The processing strategy of u-shaped obstacles which can improve the optimization ability of the algorithm by means of processing the pheromones on the path of the ants trapped in the u-shaped obstacle was proposed.Secondly,The path optimization strategy was proposed,and the optimal path found in each iteration is optimized.The optimized path is shorter and smoother.Thirdly,Uneven distribution of initial pheromone concentration which improves the algorithm's initial optimization ability by designing the corresponding formula.Fourthly,according to the path-finding characteristics in the iterative process,We Divide the whole process into three stages and design corresponding formulas to achieve the adaptive updating of pheromone heuristic factor,visibility heuristic factor and pheromone volatile factor.Fifthly,in order to reduce the robot's corner loss,transfer parameters are introduced to improve the probability transfer formula,and the concept of passability is introduced into the transfer mode selection strategy.Sixthly,the evaluation mechanism has been updated to make the choice of path more reasonable.Finally,experimental comparison shows that the improved algorithm has fast speed and strong global optimization ability.