Research On Intelligent Vehicle Location And Path Planning

On traffic roads,there are various complicated conditions.For smart cars driving on the road,these conditions will cause the intelligent driving system to be unable to accurately determine their specific location in the environment at a certain speed and plan to avoid them in time.The optimal path of the surrounding obstacles causes the automatic braking system to fail,cause accidents,and cause heavy losses of personnel and property.Therefore,it has important theoretical significance and application value for the in-depth study of smart car positioning and path planning algorithms and improving the performance of all aspects of the algorithm.The thesis focuses on the self-built smart car research,the main content includes the following aspects:(1)Establish various system models of smart vehicles,including smart vehicle software and hardware system models,kinematics and dynamics models,sensor models,etc.Through the construction of these models,we can have a deeper understanding of the smart car system,and build a real experimental platform based on these models.(2)In-depth study of the path planning algorithm,and use the grid method to build an environmental map in Matlab,simulate and verify the traditional ant colony algorithm and the A*algorithm,analyze the advantages and disadvantages of the two algorithms,and propose improved ideas.(3)Aiming at the problems of traditional ant colony algorithm in the search efficiency of path nodes is low and planning paths are not optimal,etc.,the traditional ant colony algorithm is designed for the initialization of pheromone,the initial value strategy of state transition,and the update of global pheromone.New improvement strategy,and simulation verification in Matlab.(4)Propose an effective method to improve the particle sampling effect and reduce the number of resampling.First,when solving sampling based on particle filter,the mean and variance of the posterior probability calculated by the extended Kalman filter are used as the mean and variance of the suggested density for weight calculation.Subsequently,the standard deviation of the weight and the number of effective particles are used as the evaluation of the severity of the weight degradation of the particle filter algorithm,and the number of less than the weight average is counted.When it is greater than the preset threshold,it is determined that re-sampling is required.Finally,the extended Kalman filter,the particle filter positioning algorithm and the fusion algorithm were simulated by Matlab,and the effectiveness and realtime performance of the fusion method were compared and analyzed.(5)Combining the advantages and disadvantages of the A*algorithm and the improved ant colony algorithm,an improved ant colony-A*fusion global path planning algorithm is designed.The fusion algorithm is based on the improved ant colony algorithm,and the A*algorithm is the auxiliary algorithm for dual planning of the global environment,which improves the convergence speed and search efficiency of the overall algorithm,and solves the problem that the planning algorithm is easy to fall into the local optimal value.In Matlab,the grid method is used to establish the environment map.Through comparative analysis,the feasibility of the algorithm is verified.Aiming at the problem of the unsmooth route planned by the grid map,the B-spline curve algorithm is used to optimize it to make it more in line with the actual smart car trajectory,improve the smoothness of the route,and reduce unnecessary peak inflection points.(6)A composite adaptive neural network tracking control strategy based on disturbance observer is proposed.A composite adaptive learning rate is constructed by designing a secondorder low-pass filter identification model,in which the adaptive parameters are jointly adjusted by the modeling error vector and the tracking error,which not only effectively attenuates the measurement noise but also accelerates the convergence speed of the system tracking error.The interference observer is designed to estimate the composite interference received by the system,and the observed interference value is canceled in the input channel,which improves the antiinterference ability of the system.The algorithm was verified in real vehicles through Matlab and the research group’s smart car.An optimal desired path is planned through the studied path planning algorithm.The smart car uses the neural network to control the car along the desired path to ensure that the center of the car is consistent with the planned route,so as not to deviate from the route and improve intelligence.The anti-interference ability of the car in the actual environment ensures the safe driving of the smart car.