Improvement And Application Of Local Path Planning Algorithm Based On Environmental Information

With the gradual improvement of mobile robot intelligence,it has evolved from a simple mechanical structure to an intelligent individual with complex functions,and has been widely used in both civil and military fields.As an important part of the motion planning module,the research on path planning algorithm continues to increase.In this paper,the local path planning algorithm is improved based on the consideration of vehicle kinematics constraints and computational complexity.In order to solve the problem that the traditional dynamic window algorithm(DWA)cannot effectively plan the route in a specific environment,or it is easy to fall into a local optimum,this paper proposes a VA-DWA path planning algorithm based on environmental information to achieve the goal that agents can still plan reasonable and effective paths in complex and dynamic environments.The specific research contents are as follows:(1)Aiming at the loophole of angle calculation delay in the original DWA algorithm in the course angle evaluation function,this paper takes the terminal position of discrete speed sampling track as the input of the course angle evaluation function,so that the agent can guide to the target point in time when turning to a large extent.(2)For the common straight line and "C" type trap obstacles,the traditional DWA algorithm cannot escape quickly.This paper proposes a simple and efficient in place turning strategy,which enables the agent to turn to a certain angle before the collision,so as to plan a new path to escape the local trap.(3)A new A-DWA algorithm based on adaptive evaluation function is proposed.In the dense obstacle and narrow gap obstacle environment,the local map information returned by the sensor affects the track evaluation function of the DWA algorithm in the form of adaptive evaluation function parameters,changes the proportion of the heading angle and obstacle distance evaluation function,so that the agent pays more attention to nearby obstacles,so that it can quickly pass through the dense obstacle area and narrow gap area in front of it,and shorten the overall running time.Finally,the effectiveness of A-DWA algorithm is proved by simulation under random map.(4)An obstacle avoidance strategy of VA-DWA algorithm based on virtual obstacle location prediction in dynamic environment is proposed.In view of the uncertainty of the dynamic environment,the obstacle is divided into two types of motion: straight line movement and curve movement.A kind of obstacle avoidance strategy is determined,which takes the dynamic obstacle trajectory sampling as the prior information,and obtains the position at a certain time in the future through the fitting function.By deploying the corresponding virtual obstacle on the predicted moving trajectory of the dynamic obstacle,the agent can reduce the possibility of lateral collision after crossing the dynamic obstacle.Finally,the simulation experiment under random map proves the superiority of the improvement.(5)Finally,the rationality and effectiveness of the above research contents are verified by ROS intelligent vehicle equipped with VA-DWA algorithm in this paper.