Reliable condition monitoring of transmission line fittings is essential to improve grid safety and reduce maintenance costs. However, defect detection in unmanned aerial vehicle (UAV) inspection imagery is challenging due to extreme scale variation, tiny targets such as bolts and pins that are prone to false positives and missed detections, cluttered backgrounds, and hard samples such as small objects interspersed among regular-sized objects that slow convergence. To address these issues, this study proposes an artificial intelligence-based defect detection method that combines a two-stage transfer learning strategy with image background denoising, enabling the detector to gradually learn discriminative features of transmission line fittings. On this basis, the You Only Look Once version 11 (YOLOv11) detector is improved to enhance multi-scale feature extraction, improve adaptability to small targets, and mitigate the influence of hard samples during training. Experimental results show that the proposed method achieves a precision of 88.4%, a mean Average Precision of 81.3% at an Intersection over Union threshold of 0.5 ([email protected]), and a mean Average Precision of 60.3% averaged over Intersection over Union thresholds from 0.5 to 0.95 ([email protected]:0.95), while running at 68.1 frames per second (FPS). Compared with the corresponding baseline detector, the proposed method improves these metrics by 4.7, 4.5, and 4.9 percentage points, confirming its effectiveness for transmission line fitting defect detection in complex backgrounds and its practical potential for near-real-time UAV-assisted transmission line inspection.

This paper addresses the issue of stiffness in the upper body and lack of coordination between the upper and lower body during humanoid robot walking. An improved humanoid robot reinforcement learning algorithm incorporating an LSTM framework is proposed to optimize full-body coordinated movement. Based on the Humanoid-Gym framework, a novel reward mechanism is designed, taking into account the detailed evaluation of arm movement and the collaborative control between the arms and thighs. The reinforcement learning model adopts an Actor-Critic architecture, integrating the LSTM framework into the network to enhance the feature extraction and dynamic modeling capabilities. Finally, experiments were conducted using the Hi ROBOT humanoid platform to validate the proposed model. The proposed LSTM network algorithm is compared with the original network, GRU, CNN, and other networks, demonstrating the superiority of the model. Compared with other networks, the performance improves by approximately $3.4 \%$ in terms of reward metric, and the model reaches the performance level of the original network after 40k training steps, as opposed to 60 k steps. It also maintains a fast convergence rate. Additionally, the optimized algorithm results in better gait arm swing and leg coordination, with smoother and more coordinated movement, closest to the human walking pattern.

When traditional path planning algorithms are used for path searching for Automated Guided Vehicles (AGVs) in static environments, they usually encounter the difficulties of excessive search nodes, high memory consumption, and long running time. To tackle these problems, this paper proposes a tangent point search algorithm based on quadtree grid environment modeling. The algorithm establishes a weighted graph by extracting key cell points and then uses the Dijkstra algorithm for path planning. In the scenarios of different map sizes, A* algorithm, Rapidly-exploring Random Trees (RRT) algorithm, Dijkstra algorithm, and a Dijkstra algorithm with a tangent point search (DA+TPS) were simulated and analyzed. The results show that among the four algorithms, the DA+TPS has the shortest route length and minimum search time. The comparison demonstrates that the proposed method can effectively reduce the number of search nodes, speed up the search process, and reduce redundant nodes in the path, thus reducing the number of turns for AGVs.

Improving the classification accuracy rate of motor imagery (MI) tasks, which is the crucial challenges in designing brain-computer interface (BCI) systems. A compound model of convolutional neural networks (CNN) serially connected gated recurrent unit (GRU) is proposed in this paper, which can simultaneously extract spatial-temporal information of EEG signals to enhance the accuracy in MI binary classification tasks. This method is applied to the motor imagery datasets of 16 subjects from Inner Mongolia University of Technology (IMUT). Experimental results showed that the model of CNN+GRU has a significantly higher classification accuracy than the model of CSP+SVM. More abundant fusion features extracted by CNN+GRU model, and the accuracy rate of MI features classification has been significantly improved.

The current nonlinear optimization of visual SLAM back-end has disadvantages such as slow optimization speed and poor optimization effect. To overcome these problems, this paper improves the traditional Levenberg-Marquardt algorithm (L-M) based on a framework of Bundle Adjustment (BA) nonlinear optimization. Firstly, the radius and expansion multiplier of the trust region are formulated and a threshold value is set; secondly, the trust region after each iteration is restricted with the pre-defined range for the purpose of improving nonlinear optimization; finally, a comparative analysis is conducted by setting up a comparison experiment with the traditional L-M algorithm, and It is concluded that the improved L-M algorithm can reduce the number of iterations by 16 times and time performance was reduced by an average of 62.30%, which shorens the optimization time, improves the optimization efficiency and has better robustness.

Laser slam usually needs to complete a back-end graph optimization at a fast speed in some specific scenes, such as sharp turns, fast motion, and limited calculation time. Aiming at these problems, this paper proposed a 2D laser slam back-end graph optimization combined with Cholesky decomposition to accelerate a linear solution process and further to achieve a purpose of accelerating graph optimization. In MATLAB simulation experiments, the rate of 2D laser slam back-end graph optimization combined with Cholesky decomposition increased 24%, compared to that of the traditional method without Cholesky decomposition. The result verified the effectiveness of the improved method.