Research On Collision Avoidance Of Unmanned Vehicles Under Specific Path Of Airport

Unmanned vehicles are the heart academia of artificial intelligence research,and have broad application prospects in future intelligent transportation systems.In the real environment,unmanned vehicles are affected by random dynamic obstacles and complex weather,and are prone to take place unsafe situations,such as congestion and collision.Therefore,collision avoidance is the core technology of driverless vehicles.Under the specific path of the airport corridor bridge,this paper studies the problem of path planning problem at the control decisionmaking level,and image de-fogging and de-raining at the environmental perception level.There are many traffic intersections and obstructions under the airport bridge,therefore,at the decision-making level,to optimize the path planning duration and smooth planning path for dynamic collision avoidance of unmanned vehicle,a fusion algorithm for path planning is proposed.This algorithm introduces a dynamic window approach based on a Bidirectional Rapidly-exploring Random Trees to achieve collision avoidance for unmanned vehicles under dynamic constraints.Simulation results show that the planned trajectory is more secure and more effective for dynamic collision avoidance of unmanned vehicles under airport bridge,being able to conduct collision avoidance path planning both rapidly and smoothly.In this paper,the collision avoidance research of the control decision-making layer is based on data preprocessing of the environmental perception level;Therefore,at the environmental perception level,rain and fog weather severely reduces the ability to sense the environment of unmanned driving.To eliminate halo artifacts and improve the image visibility,a manifold particle filtering algorithm is proposed.Based on the dehazing algorithm of the dark channel prior,the proposed technique uses manifold particle filter to optimize the transmission and achieve accurate transmission estimation.Finally,the image brightness is adjusted to restore a clear fog-free image.Experimental results demonstrate that the proposed algorithm is capable of removing halo artifacts and restoring the visibility of foggy image.Rain is often accompanied by the generation of fog.To solve rain marks and unclear problems in image,this paper proposes a method that optimizes the attentive generative adversarial network,combines the gaussian model with the generative adversarial network to remove the background interference.At the same time,the defogging module of manifold particle filtering is added to the attentive generative adversarial network.The experimental results show that compared with the existing rain-removal algorithm,the proposed algorithm can remove the rain marks and fog in image effectively,and improves clarity of the image significantly.The data processing of environment perception is realized by the algorithm of removing rain and fog,which lays a good foundation for the path avoidance of driverless vehicle.