With the rapid development of computer vision and deep learning, researchers have begun to focus on the semantic characteristics of traditional Simultaneous Localization And Mapping in three-Dimensional scenes. The point cloud map generated by the traditional simultaneous localization and mapping method takes up considerable storage space and cannot extract semantic information from the scene, which cannot meet the requirements of intelligent robot navigation and high-level semantic understanding. To solve this problem, this paper proposes a semantic information fusion OctoMap method. First, the color and depth images obtained from RGB-D by ORB-SLAM2 are used to locate the camera. Second, the Convolutional Block Attention Module-Pyramid Scene Parsing Network is introduced to segment the input RGB image semantically to improve the segmentation accuracy and obtain high-level semantic information in the environment. Then, a semantic fusion algorithm based on Bayesian fusion is introduced to fuse multiview semantic information. Finally, the generated semantic point cloud is inserted into OctoMap, and its octree data structure is used to compress the storage space. Experimental results based on the ADE20K dataset show that, compared with Pyramid Scene Parsing Network, Convolutional Block Attention Module-Pyramid Scene Parsing Network improves Mean Pixel Accuracy by 2.55%, and Mean Intersection over Union by 1.88%. Experimental results based on the TUM dataset show that the proposed method greatly reduces storage space and achieves the effect of voxels by voxel dense semantic mapping compared with point clouds and a traditional OctoMap.

This paper proposes a dual-loop adaptive decoupling control method based on single neuron PID controller (DADC-SNPID), balancing the single wheeled robot. A unique mechanical and control hardware structure of single wheeled robot is designed and its simplified mathematical model is established using the Newton-Euler equations according to the actual parameters of the designed robot. A new balance and motion control method is proposed as dual-loop adaptive decoupling control method based on single neuron PID controller to control the robot. Results of simulation and physical experiments are conducted to illustrate effectiveness of the designed robot system and its controller under certain conditions.

We introduce a method for modeling cancer diagnosis at the molecular level using a Chinese microarray gastric cancer dataset. The method combines an artificial neural network with a decision tree that is intended to precede standard techniques, such as classification, and enhance their performance and ability to detect cancer genes. First, we used the relief algorithm to select the featured genes that could unravel cancer characteristics out of high dimensional data. Then, an artificial neural network was employed to find the biomarker subsets with the best classification performance for distinguishing cancerous tissues and their counterparts. Next a decision tree expression was used to extract rules subsets from these biomarker sets. Rules induced from the best performance decision tree, in which the branches denote the level of gene expression, were interpreted as a diagnostic model by using previous biological knowledge. Finally, we obtained a gastric cancer diagnosis model for Chinese patients. The results show that using the Chinese gastric biomarker genes with the diagnostic model provides more instruction in biological experiments and clinical diagnosis reference than previous methods.