In order to effectively solve the overfitting problem in deep Convolutional Neural Network (CNN), a model prediction averaging method based on Dropout improved CNN was proposed. Firstly, Dropout was employed in the pooling layers to sparse the unit values of pooling layers in the training phase. Then, in the testing phase, the probability of selecting unit value according to pooling layer Dropout was multiplied by the probability of each unit value in the pooling area as a double probability. Finally, the proposed double-probability weighted model averaging method was applied to the testing phase, so that the sparse effect of the pooling layer Dropout in the training phase was able to be better reflected on the pooling layer in the testing phase, thus achieving the low testing error as training result. The testing error rates of the proposed method in the given size network on MNIST and CIFAR-10 data sets were 0.31% and 11.23% respectively. The experimental results show that the improved method has lower error rate than Prob. weighted pooling and Stochastic Pooling method with only the impact of pooling layer on the results considered. It can be seen that the pooling layer Dropout makes the model more generalized and the pooling unit value is helpful for model generalization and can effectively avoid overfitting.

To overcome the randomness of RRT-Connect and slow convergence of B-RRT ^*(asymptotically-optimal Bidirectional Rapidly-exploring Random Tree) in path generation, an efficient path planning algorithm based on B-RRT ^*, abbreviated as EB-RRT ^*, was proposed. Firstly, an intelligent sampling function was intriduced to achieve more directional expansion of random tree, which could improve the smoothness of path and reduce the seek time. A rapidly-exploring strategy was also added in EB-RRT ^* by which RRT-Connect exploration mode was adopted to ensure rapidly expanding in the free space and improved asymptotically-optimal Rapidly-exploring Random Tree (RRT ^*) algorithm was adopted to prevent trapped in local optimum in the obstacle space. Finally, EB-RRT ^* algorithm was compared with Rapidly-exploring Random Tree (RRT), RRT-Connect, RRT ^* and B-RRT ^* algorithms. The simulation results show that the improved algorithm is superior to other algorithms in the efficiency and smoothness of path planning. It reduced the path planning time by 68.3% and the number of iterations by 48.6% compared with B-RRT ^* algorithm.