In complicated environments, which include dynamic and narrow areas, the path planning of Autonomous Mobile Robots (AMRs) encounters challenges, like slow model convergence and limited representational capabilities, often resulting in the robot taking longer, less efficient paths or even colliding with obstacles. To tackle these challenges, the Gated Attention Prioritized Experience Replay Soft Actor-Critic (GAP_ SAC) algorithm is proposed. Key improvements include expanding the state space for better perception, designing a dynamic heuristic reward function to more effectively guide the AMR in achieving its path planning objectives and integrating Prioritized Experience Replay (PER) to improve sample efficiency and accelerate convergence. Additionally, a gated attention mechanism is also introduced to focus on critical environmental features, enhancing the models’ perception capability. Comparative experiments validate that the proposed GAP_SAC algorithm outperforms TD3, SAC and SAC’s variant, demonstrating superior robustness and generalization in complicated environments.

In this paper, we design an object sorting robot system, which is based on robot operating system distributed processing framework. This system can communicate with human beings; can percept the 3-D environment by Kinect sensor; has the ability of reasoning; can transfer the natural language intention to machine instruction to control the movement of manipulator. In particular, in order to improve the intelligence and usability of our robot, we propose a human-robot-environment interactive reasoning mechanism. In our method, a “dialogue and 3-D scene interaction module” is added into the traditional case-based reasoning-belief-desire-intention mechanism. Our proposed mechanism not only realizes the traditional function of map matching but also achieves the function of desire analysis and guidance. When the user's desire is incomplete and/or mismatched with the actual scene, our robot will take the initiative to guide users through dialogue, and the user's input information will be used to replenish the user's desire. Experimental results prove the advantages of our mechanism.

With the development of the technologies such as computer vision, natural language interaction, process control technology, and sensor technology, the discussion of automatic sorting system of robot hits a hotspot in the robot community. This paper presents an automatic sorting system for industrial robot with the technologies of 3D visual perception, natural language interaction and automatic programming. Therein, a ‘rule-scene’ matching and interaction algorithm is proposed to combine all these modules together. By utilising our algorithm, robot can interact with human beings according to the real-time actual three-dimensional scene information and can guide users to give correct rules through speech when the rule is invalid. After getting the correct rules, robot can sort the object automatically using the automatic programming and execution algorithm designed in this study. In the experimental section, the designed system is applied to a fruit-sorting scene, which proves the effectiveness and practicability of the system.

At present, the learning-based down-sampling method has achieved good result by using the loss of subsequent tasks to optimize the process of sampling points selection. However, for the current methods there still has two problems, on the one hand, the sampled points may not be in the original point cloud, and the sampled generated points of the sample need to be matched with the original point cloud (re-matching problem), on the other hand, the number of points obtained after matching may be insufficient for the sampling demand, which need to be supplemented by Farthest Point Sampling, nearest neighbor point, or other methods (re-supplement problem). In order to solve the above problems of re-matching and re-supplement in the learning-based sampling method, we propose a new sampling network which based on high score selection strategy. In our method, LSTM (Long Short Term Memory) is used as the feature extraction method to obtain the relational weight which relative to the subsequent tasks. The attention mechanism is used to select the high score points. The experimental results show that the performances of our network (We dubbed it as LA-Net) are better than other sampling methods and achieve the state-of-the-art in the unified benchmark. Detailed ablation studies have also been carried out, which provably affirm the advantage of each component we proposed. Finally, the methods in this paper are applied to a real backbone network to optimize its performance, and real robot grasping experiments are designed to realize automatic robot grasping of multi-target objects in real scenes.

Defect detection in printed circuit boards (PCBs) is essential for ensuring the stability and reliability of electronic equipment. However, due to the compact structures and complex backgrounds of PCBs, accurately detecting small defects while maintaining strong model generalization remains a significant challenge. To address this issue, this paper proposes YOLO-Efficient-Multiscale-Adaptive-Channel (YOLO-EMAC), a novel framework built upon an enhanced YOLOv12 architecture, designed to improve the accuracy, efficiency, and generalization of small defect detection. Specifically, to mitigate the loss of shallow features caused by extensive downsampling and convolutions, the Multi-scale and Multi-branch Cross-scale Convolutional Fusion (M2C2f) module is introduced, integrating Window-based and Multi-Head Self-Attention (WindowMHSA) with dynamic transformation layers to enhance feature extraction for small objects. Furthermore, to address feature degradation during upsampling, a Efficient-Adaptive-Multibranch-Channel (EAMC) module is developed to adaptively strengthen key feature. Extensive experiments conducted on the PCB Defect Dataset, PCB Surface Defect Dataset, and DeepPCB dataset demonstrate that YOLO-EMAC achieves SOTA detection performance. The source code, pretrained models and datasets are made publicly available on GitHub: https://github.com/wuhan66/YOLO-EMAC/tree/main.

In this paper, an intelligent water shooting robot system for situations of carrier shake and target movement is designed, which uses a 2 DOF (degree of freedom) robot as an actuator, a photoelectric camera to detect and track the desired target, and a gyroscope to keep the robot’s body stable when it is mounted on the motion carriers. Particularly, for the accurate shooting of the designed system, an online tuning model of the water jet landing point based on the back-propagation algorithm was proposed. The model has two stages. In the first stage, the polyfit function of Matlab is used to fit a model that satisfies the law of jet motion in ideal conditions without interference. In the second stage, the model uses the back-propagation algorithm to update the parameters online according to the visual feedback of the landing point position. The model established by this method can dynamically eliminate the interference of external factors and realize precise on-target shooting. The simulation results show that the model can dynamically adjust the parameters according to the state relationship between the landing point and the desired target, which keeps the predicted pitch angle error within 0.1°. In the test on the actual platform, when the landing point is 0.5 m away from the position of the desired target, the model only needs 0.3 s to adjust the water jet to hit the target. Compared to the state-of-the-art method, GA-BP (genetic algorithm-back-propagation), the proposed method’s predicted pitch angle error is within 0.1 degree with 1/4 model parameters, while costing 1/7 forward propagation time and 1/200 back-propagation calculation time.