Research On Trajectory Planning For UAV

With the continuous development of the artificial intelligence and the computer technology, the UAV(Unmanned Aerial Vehicle) trajectory planning has become a major research contents of task planning system. But in actual application, the flying environment is quite complex, and many constraints may appears. Therefore, the key problem of trajectory planning is to establish good planning environment and effective planning algorithm.Firstly, background and related technologies of the UAV trajectory planning have been introduced in this thesis,several methods of acquiring flying altitude data in flight regional environment and several trajectory planning algorithm simply are introduced too. Before studing the algorithm , flying environment mathematical model should be established.This paper introduces several elevation data access method,for processing the elevation data, choose two-dimensional three convolution interpolation method. Using RBF neural network to establish the mathematical model of peaks and threats within the scope of the flight. After fast data fusion,equivalent digital elevation figure has been established.In order to reduce the amount of constraint of trajectory planning,the constraint conditions merged into digital elevation,the minimum threat surface was built up,unmanned Aerial Vehicle trajectory points of Unmanned Aerial Vehicle is in the threat surface. Established the mathematical model and trajectory planning model of constraints after finished environment model, On the basis of the model proposed, The paper puts forward the improved mathematical model through adding random points, Not only reduce the computation, but also can avoid route through the threat of regional problems. Trajectory planning divided into online and off-line, Using A* algorithm, genetic algorithm and genetic simulated annealing algorithm when execute the off-line trajectory planning.In the application of A* algorithm, use of motor vehicle constraints again to reduce the search space,the point in the searching space of making the index function smallest is the flying node, repeat this process, the whole trajectory is available. In the application of the standard genetic algorithm, the local extremum problems appeared, for the problem using simulated annealing algorithm was improved.On-line flight stage, unmanned aerial vehicle fly according to the planned reference route.Flight process real-time detection front, If there is a new threat, the threat data will be merged quickly into digital elevation, Using the optimal control theory ,planning a new trajectory to avoid threat real-time, Then go back to the original route. Finally, using the method combined A*algorithm with genetic algorithm, one entirely new routes is appeared. The simulation results are good.