Path-planning research has been the key to mobile-robot-navigation technology. However, traditional path-planning algorithms have some shortcomings. To solve these problems, this paper proposes a fusion algorithm that combines the kinematical constrained A* algorithm with the Dynamic Window Approach (DWA) algorithm. The kinematical constrained A* algorithm can plan the global path, and then the DWA algorithm can plan the local path under the global path’s guidance. Firstly, combined with robot kinematics, we improve the node-expansion method and heuristic-function model of the A* algorithm, which improves the search efficiency, reduces the number of path bends and lowers the computational cost so that the path generated by the A* algorithm better meets the needs of robot motion. Secondly, we optimize the trajectory-evaluation function of the DWA algorithm so that the local paths planned by the DWA algorithm are smoother and more coherent, which is easier for robot-motion execution. Finally, we extract the key nodes from the global path planned by the A* algorithm to guide the DWA algorithm for local path planning and dynamic-obstacle avoidance and to make the local path closer to the global path. Through simulation and practical experiments, the effectiveness of the fusion algorithm proposed in this paper is verified.

In this paper, we propose a localization method applicable to 3D LiDAR by improving the LiDAR localization algorithm, such as AMCL (Adaptive Monte Carlo Localization). The method utilizes multiple sensing information, including 3D LiDAR, IMU and the odometer, and can be used without GNSS. Firstly, the wheel speed odometer and IMU data of the mobile robot are multi-source fused by EKF (Extended Kalman Filter), and the sensor data obtained after multi-source fusion are used as the motion model to participate in the positional prediction of the particle set in AMCL to obtain the initial positioning information of the mobile robot. Then, the position pose difference values output by AMCL at adjacent moments are substituted into the PL-ICP algorithm as the initial position pose transformation matrix, and the 3D laser point cloud is aligned with the nonlinear system using the PL-ICP algorithm. The three-dimensional laser odometer is obtained by LM (Levenberg--Marquard) iterative solution in the PL-ICP algorithm. Finally, the initial position pose output by AMCL is corrected by the three-dimensional laser odometer, and the AMCL particles are weighted and sampled to output the final positioning result of the mobile robot. Through simulation and practical experiments, it is verified that the improved AMCL algorithm has higher positioning accuracy and stability compared to the AMCL algorithm.

In order to meet the needs of intelligent perception of the driving environment, a point cloud registering method based on 3D NDT-ICP algorithm is proposed to improve the modeling accuracy of tunneling roadway environments. Firstly, Voxel Grid filtering method is used to preprocess the point cloud of tunneling roadways to maintain the overall structure of the point cloud and reduce the number of point clouds. After that, the 3D NDT algorithm is used to solve the coordinate transformation of the point cloud in the tunneling roadway and the cell resolution of the algorithm is optimized according to the environmental features of the tunneling roadway. Finally, a kd-tree is introduced into the ICP algorithm for point pair search, and the Gauss–Newton method is used to optimize the solution of nonlinear objective function of the algorithm to complete accurate registering of tunneling roadway point clouds. The experimental results show that the 3D NDT algorithm can meet the resolution requirement when the cell resolution is set to 0.5 m under the condition of processing the point cloud with the environmental features of tunneling roadways. At this time, the registering time is the shortest. Compared with the NDT algorithm, ICP algorithm and traditional 3D NDT-ICP algorithm, the registering speed of the 3D NDT-ICP algorithm proposed in this paper is obviously improved and the registering error is smaller.

Real-time dense 3D reconstruction is one of the major challenges in computer vision and robotics. In this paper, we propose a real-time 3D reconstruction model with metric-scale, including a direct visual-inertial odometry with stereo cameras and a deep multi-view stereo network. Aiming at the scale uncertainty of dense map constructed by monocular camera, we designed a direct stereo visual-inertial odometry(DSVIO). The odometry combines static stereo optimization with direct visual-inertial odometry, using left-right images to initialize the depth of feature points, which can significantly improve the accuracy of 6-DoF(Degree of Freedom) pose and metric-scale in the active window. In the aspect of depth estimation, the minimizing photometric re-projection loss(MPRP) proposed by us can integrate the common view-points for depth estimation under different view to improve the performance of CVA-MVSNet. Finally, the predicted depth map is fused into the truncated signed distance function (TSDF) voxel volume. The experiment shows that the pose estimation of our visual odometer has state-of-the-art(SOTA) performance when the trajectory is smooth and low jitter. In the case of fast jitter, our method is still superior to the monocular visual-internal odometry of ORB-SLAM3, but slightly inferior to the stereo visual-internal odometry of ORB-SLAM3.In the experiment of depth estimation, MPRP effectively improves the performance of CVA-MVSNet, and all evaluation indicators were superior to the original method. Moreover, our method had good performance in real-time 3D reconstruction with metric-scale.

Due to the problem of poor recognition of data with deep fault attribute in the case of traditional superficial network under semisupervised and weak labeling, a deep belief network (DBN) was proposed for deep fault detection. Due to the problems of deep belief network (DBN) network structure and training parameter selection, a stochastic adaptive particle swarm optimization (RSAPSO) algorithm was proposed in this study to optimize the DBN. A stochastic criterion was proposed in this method to make the particles jump out of the original position search with a certain probability and reduce the probability of falling into the local optimum. The RSAPSO-DBN method used sample data to train the DBN and used the final diagnostic error rate to construct the fitness value function of the particle swarm algorithm. By comparing the minimum fitness value of each particle to determine the advantages and disadvantages of the model, the corresponding minimum fitness value was selected. Using the number of network nodes, learning rate, and momentum parameters, the optimal DBN classifier was generated for fault diagnosis. Finally, the validity of the method was verified by bearing data from Case Western Reserve University in the United States and data collected in the laboratory. Comparing BP (BP neural network), support vector machine, and heterogeneous particle swarm optimization DBN methods, the proposed method demonstrated the highest recognition rates of 87.75% and 93.75%. This proves that the proposed method possesses universality in fault diagnosis and provides new ideas for data identification with different fault depth attributes.

The unevenness and darkness of lighting in real environments are one of the main issues that affect the practical application of visual SLAM. To address this issue, we propose a visual SLAM algorithm based on low-light enhancement networks. Firstly, we designed a lightweight low-light enhancement network based on deformable power curves, DPCE-Net, to adjust the pixels‘ dynamic range of a low-light image. The curve is derived from power function, which considers the range, monotonicity and differentiability of pixel values, and has a high dynamic range. DPCE-Net is trained and evaluated without reference images, effectively expanding the application scenarios of low-light enhancement. Then, combine the DPCE-Net with ORB-SLAM2 to construct a low-light enhancement SLAM(DPCE-SLAM). In the low-light image enhancement experiment, we compared with the state-of-the-art method, Enlighten-GAN and Zero-DEC, achieving the good performance in SSIM, MSE, PSNR, and NIQE indicators. In state estimation evaluation, DPCE-SLAM can effectively increase the number of correctly matched feature points in dark environments, while also improving the robustness of state estimation.