To overcome low pathfinding efficiency and slow convergence speed of Rapid-exploring Random Tree star(RRT ^*) in high-dimensional and complex environment, an Unmanned Aerial Vehicle(UAV) path planning algorithm with self-adjusting sampling space based on RRT ^* named Adjust Sampling space-RRT ^*(AS-RRT ^*) was proposed. In this algorithm, by adjusting the sampling space adaptively, the tree was guided to grow more efficiently, which was realized through three strategies including:biased sampling, node selection and node learning. Firstly, the light and dark areas in the sampling space were defined to performing biased sampling, and the probability weights of the light and dark areas were determined by the current expansion failure rate, so as to ensure that the algorithm was both exploratory and directional when searching for the initial path. Then, once the initial path was found,the nodes were periodically filter,and the high-quality nodes were used as learning samples to generate the new sampling distribution, the lowest-quality nodes were replaced by new nodes after the algorithm reaching the maximum number of nodes. Simulation experiments for comparison were conducted in multiple types of environments. The results show that the proposed algorithm improves the inherent randomness of the sampling algorithm to a certain extent, and compared with the traditional RRT ^* algorithms, it has less pathfinding time used in the same environment, lower cost path generated in the same time, and the improvements are more obvious in three-dimensional space.

According to the feature of high-dimensional data, the number of variables is usually larger than the sample size and the data are often heterogeneous, a robust and effective feature selection method was proposed by using the dimensional reduction technique of variable selection and the modal regression based estimation method. The estimation algorithm was given by using Local Quadratic Algorithm (LQA) and Expectation-Maximum (EM) algorithm, and the selection method of the parameter adjustment was also discussed. Data analysis of the simulation shows that the proposed method is overall better than the least square and median regression based regularized method. Compared with the existing methods, the proposed method has higher prediction ability and stronger robustness especially for the non-normal error distribution.