This paper presents a maritime path planning algorithm for autonomous surface vehicles (ASVs). A improved RRT algorithm for generating sample nodes is proposed. The HOCBF-RRT algorithm leverages high order control barrier functions (HOCBFs) to take into account the dynamics of the ASV and produce a path that prioritizes safety. In addition, to complete the path optimization, combined with control Lyapunov functions (CLFs), the expansion to the target point is accelerated and the planning time is reduced. In simulation experiments, it is observed that the HOCBF-RRT algorithm and HOCBF-CLF-RRT algorithm offer improvements over traditional RRT algorithms. Specifically, the HOCBF-RRT algorithm is found to enhance the smoothness of paths, while the HOCBF-CLF-RRT algorithm effectively optimizes the expansion of the random tree towards the target point.
In recent years, the conservation of water resources has attracted widespread attention. The development and application of water surface robots can achieve efficient cleaning of floating waste. However, limited to the small size of floating waste on the water surface, its detection remains a great challenge in the field of object detection. Existing object detection algorithms cannot perform well, such as YOLO (You Only Look Once), SSD (Single-Shot Detector), and Faster R-CNN. In the past two years, diffusion-based networks have shown powerful capabilities in object detection. In this paper, we decouple the position and size regressions of detection boxes, to propose a novel decoupled diffusion network for detecting the floating waste in images. To further promote the detection accuracy of floating waste, we design a new box renewal strategy to obtain desired boxes during the inference stage. To evaluate the performance of the proposed methods, we test the decoupled diffusion network on a public dataset and verify the superiority compared with other object detection methods.
Tanker ship is an important transportation equipment for petroleum. Due to the strong permeability and solubility of refined oil product, the high-quality inner wall special coating of tanker ships needs to be carried out regularly, which relies on manual operation with poor working conditions in confined space, high labor intensity, long cycle and fluctuating coating quality. This paper designs a lightweight rigid-flexible robot with a cable-driven parallel robot (CDPR) and a symmetric mechanical arm in series. The workspace analysis of the CDPR is completed based on kinematic and static modeling. The size of the symmetric mechanical arm and the terminal reachability are analyzed considering the interference. The results show that the proposed rigid-flexible cable robot can cover all the areas to be coated, which provides a new automatic solution for the inner wall special coating of tanker ships.
With the rapid development and wide application of industrial robots (IRs), it inevitably brings huge energy consumption (EC), which has become an important part of manufacturing EC. Therefore, the EC optimization of IRs has become the key to the green transformation and upgrading of the manufacturing industry, and it is of great significance for the realization of "carbon neutrality" and "carbon peaking". Therefore, this paper focuses on the energy evaluation and optimization of IR, and realizes the power and EC evaluation and motion parameter optimization of its trajectories based on data-driven method. First, the convolutional neural network (CNN) and Transformer models are combined to build the energy model of IR, and then accurate modeling of its power and EC is realized based on the deep learning algorithms. Based on the above energy model, the exhaustive method and genetic algorithm (GA) are used to find the optimal motion parameters and obtain the optimal trajectory with the least EC. Finally, the experimental results show that the proposed method can achieve more than 98% and 99% of the power and EC modeling of IR, respectively. The optimization of the trajectory motion parameters of IR is realized through exhaustive method and GA, and the maximum optimization potential on the test dataset can reach 52.77%, which verifies the effectiveness and accuracy of the proposed method.
In order to satisfy the demand for large planar workspace and lightweight in ship side painting, we designed a cable-driven parallel robot (CDPR) with parallel cables powered by a common drive unit considering easy deployment and structural simplicity. Different parallel cables and the effects of cable numbers are discussed. Based on the analysis, typical configurations of parallel cables are proposed, and the configuration of the CDPR for ship side painting is deduced, which has two translational degrees of freedom (DOF), and all three rotational DOFs are constrained. Kinematic and static modeling are executed, and dimensional optimization is carried out with the goal of maximizing the workspace. Workspace coverage rate of 88% is achieved with a 30m×20m vertical area.
This paper deals with the assignment problem of multiple robot with the rectangular spaying tasks. Without pointing to the starting points of each task, the upper left vertex, the upper right vertex, the lower left vertex and the lower right vertex are selected by the genetic algorithm. The ergodic-based genetic algorithm is designed to achieve the shortest time and the lowest path cost. The improved mutation operator is set to accelerate the convergence process and improve the practicability of the proposed algorithm. Compared with the strategy of market-based algorithm, the genetic algorithm reduces the average time cost by 16.98% and distance costs by 9.05%, respectively.
The paper proposes a path planning method based on a simplified visibility graph (SVG) and support vector machine (SVM). Firstly, the centroid point set of polygon obstacles is used to construct a Delaunay triangulation network. Then, all neighbors of each obstacle are obtained from the Delaunay triangulation network to construct neighbor pairs. A global environmental SVG is then constructed by simplifying the visibility relationships between vertices of neighbor obstacle pairs. Next, the start and goal points are maintained in the SVG and the path is planned using Dijkstra’s algorithm and pruning method. Additionally, an elliptical window is generated based on the globally planned path. The obstacles and the boundary of the ellipse are classified into positive and negative samples, respectively. A safe and smooth path is obtained by training the SVM based on the radial basis function (RBF). The simulation experiments show that, compared with traditional methods, the proposed method based on the Delaunay triangulation reduces the number of visual edges by more than 40% and reduces cost time by over 50%. What’s more, the optimized path obtained through the SVM is sufficiently smooth and safe.
With the continuous development of tennis tournaments, the Tennis Ball Boy (TBB) mechanism has been gradually improved, the Eagle-Eye Vision System has been maturely applied in tennis tournaments, and TBB has been moving towards intelligence. In this paper, the concept of an intelligent TBB robot system is proposed, and a TBB robot based on a quadrotor Unmanned Aerial Vehicle (UAV) is developed to address the problems of long working hours, high intensity and the safety hazards of standing for TBB in tennis tournaments. Based on the modular design method, the intelligent TBB robot includes robot hardware design, visual tennis ball object detection, visual localization and navigation, and overall robot control system design. The tennis ball clamping mechanism is designed based on a symmetric single Degree-Of-Freedom (DOF) pawl mechanism. The tennis ball object detection system is designed with the YOLOX deep learning object detection framework, and this model is deployed through the OpenVINO platform, which enables the robot to detect tennis balls in real-time. The visual localization system based on the VINS-FUSION framework is constructed with a binocular vision fusion IMU. On this basis, a SE(3) positional controller, a 3D A* path planning algorithm and Minimum Snap trajectory optimization are developed to enable the robot to have navigation and obstacle avoidance functions. An advanced decision control system based on the Finite State Machines (FSM) is designed for the Robot Operating System (ROS) platform. Finally, the experiment verifies the reliability of each module of the robot, and this paper can provide a basis for further research.
The use of monocular cameras for ego-motion estimation in autonomous driving is a fundamental technique with significant potential for development. Recently, unsupervised CNNs based schemes have rapidly advanced due to their label-free advantages. However, traditional CNNs lack the ability to capture global dependencies, which are essential for sequential tasks like pose, depth, and optical flow estimation. In this paper, according to the characteristics of these three-independent tasks, we designed different network frameworks fully based on the transformer separately to improve performance by taking advantage of the global dependencies of different tasks. After unsupervised training, the model can independently predict the pose, depth, and optical flow results for the vehicle, solely based on a monocular camera. The experimental results conducted on the KITTI demonstrate that the proposed method obtain satisfactory results in all three tasks.
This paper proposes a novel distributed task allocation method based on improved contract network protocol for collaborative task allocation of reconnaissance and strike tasks on heterogeneous unmanned aerial platforms. Firstly, considering the heterogeneity and endurance of unmanned aerial platforms, as well as the time-sensitive characteristics of tasks, a task allocation model was established with the goal of maximizing global benefits. In the proposed method, a single unmanned platform satisfies the constraints of the task allocation model by designing the encoding and decoding methods of the existing sparrow search algorithm, and achieves the selection of its local optimal task execution path (similar to the single traveling salesman problem); By sequentially auctioning a single task, conflicts between different unmanned platforms are resolved and global benefits are improved. The bidding method for bidders, task award rules for auctioneers to select winning bidders, and judgment mechanisms for auction termination are designed. The numerical simulation results validate the feasibility of the proposed method and its superiority in global benefits compared to the contract network protocol.
The unmanned aerial vehicle (UAV) requires the control system with fast response speed, high accuracy and strong robustness. Due to the strong coupling and nonlinearity of UAV system, it is difficult to build the accurate system model, which makes it difficult for traditional linear control methods to achieve accurate and stable control effects. This paper proposes an intelligent attitude control method for high-speed UAV. Firstly, the nonlinear attitude motion model of UAV is established and a new intelligent attitude controller structure is built based on the Deep Deterministic Policy Gradient (DDPG) algorithm. Secondly, the DDPG attitude controller is trained by the simulation training system. The control simulation results show that the proposed intelligent control system and controller can meet the requirements of attitude adjustment and attitude stabilization, the control accuracy error is less than 0.02 degree. Finally, considering fixed mutation interference, Gaussian noise and 50% deviation of aerodynamic parameters disturbance, the control accuracy error is still less than 0.1 degree during the interference test. The stability analysis shows that the intelligent controller has strong robustness and generalization ability.
Surface material classification (SMC) is a critical component of robot perception system for understanding the external environment. However, traditional SMC methods that rely solely on visual information are susceptible to the effects of lighting and distance, and have limitations in classification accuracy and robustness. Drawing inspiration from the human perception system, we propose a Trimodal Fusion Neural Network (TMFNN) based on multiloss ensemble. Our model first extracts visual, auditory, and haptic features using CNN and GRU, and then efficiently fuses them using self-attention mechanism. We also employ an improved multiloss ensemble to enhance model performance. Our classification results on the TUM Dataset reveal that the proposed method significantly outperforms state-of-the-art methods, achieving the highest classification accuracy of 96.56%. Furthermore, we experimentally demonstrate that our proposed multimodal fusion and multiloss ensemble approaches can significantly enhance model accuracy. Overall, our TMFNN is capable of handling more complex SMC tasks and has considerable promise for improving robot perception in a variety of environments.
Aiming at the problem of emergency detection and search in national marine related professional fields, a Monte-Carlo method based multiple Air-Water Amphibious Robots collaborative detection method for underwater targets is studied in this paper. Firstly, a multiple Air-Water Amphibious Robots cooperative detection formation was established, and the overlapping shadow areas of different arrays were explored, and conditions were set for the search area and the Air-Water Amphibious Robots cooperative detection formation. After simulation research, parameters of each multiple Air-Water Amphibious Robots were selected and assigned, and then a set of simulation process was designed to solve the problem. Finally, the proposed method is used to simulate the relationship between detection probability and search time for 8 formations. Through the analysis of simulation results, it is concluded that the corresponding formation can be adopted to detect and track different tasks and different stages of tasks to achieve the best detection effect.
In this paper, a model-free reinforcement learning(RL) method of training a nonlinear attitude controller of a quadrotor is proposed. For the problem that the attitude controller is uncontrolled when trained by RL directly, the proposed method utilizes an expert to provide the prior information, i.e. the action’s judgement and suggestion, to guide the updating process. For the problem that the policy falls in local optima by the limitation of the expert, the proposed method maximize the entropy of the strategy to increase the exploratory behavior of the nonlinear attitude controller approximator. Furthermore, We employ the Proximal policy optimization algorithm (PPO) as the RL model and PID algorithm as the expert model to approach an exact attitude controller of a quadrotor based on the proposed method. Finally, the simulations experiments has been conducted to verify that our proposed method can train a true nonlinear attitude controller which has a better performance than the expert.
The integrated calibration method of micro-nano satellite attitude benchmark is studied. Considering the magnetic effects and residual magnetism caused by the complex magnetic environment of the satellite, an accurate measurement model for magnetometers was adopted. By using the relationship between the measured and theoretical values of the geomagnetic field, the residual magnetism and magnetic sensitivity coefficients in the magnetometer measurement model were calibrated, through the least square method of uncertain parameters of the calibration. After modeling the installation error of the solar sensor, the calibration of the solar sensor is completed, so as to realize the integrated calibration of the attitude benchmark of the micro-nano satellite. Finally, this article conducted ground experiments using telemetry data from an in orbit satellite. The experimental results showed that the proposed method reduced sensor measurement errors, improved sensor benchmark consistency, and improved attitude determination accuracy from around 10 degrees to better than 0.2 degrees, verifying the effectiveness of the proposed method.
Cable manipulator can play a lifting role in the dock stacking, mineral mining, marine fishing and other fields, which has a wide range of applications. However, the cable elasticity itself will affect the accuracy of the cable force control, which in turn affects the positional control accuracy of the end-effector. To solve the problem of control accuracy caused by simplifying the cable to a linear model without considering the cable elasticity, this paper adopts the model of simplifying the cable to a linear spring plus a lightweight connecting rod. The elasticity of the linear spring, i.e., the cable tension is derived through the static equilibrium equation. And a dynamics simulation analysis of the three-cable three-degree-of-freedom parallel manipulator is conducted. The end position and its derivative as the flat output of the system is selected and the linear process of nonlinear control theory is used to prove the differential flatness of the system. Through theoretical analysis and example simulation, the difference of the requested tension force when the cable elasticity is considered or not is compared and analyzed. It is shown that the proposed dynamics analysis of the cable parallel manipulator with cable elasticity has influence on the moment control of the whole manipulator. It is verified that the magnitude of the cable elasticity is ultimately affected by the change of the cable acceleration. And this dynamics analysis can improve the control process caused by the high speed change of the end-effector to analyze the effect of imprecise force control brought by the high speed change of cable elasticity of the cable, It provides a more accurate theoretical basis for establishing a more accurate theoretical model of dynamics
This paper puts forward a novel parallel mechanism with multiple driving modes to address the inherent limitations of workspace and singular configurations in single-driven parallel mechanisms. Taking the planar 6R parallel mechanism as an example, we conduct numerical and simulation-based studies to demonstrate the superior kinematic performance of the multi-drive mode parallel mechanism. The analytical process involved initial investigation and characterization of the mechanism, development of prototype, establishment of inverse kinematics model and introduction of local transmission index. Motion/force transmission indices under both single driving mode and multiple driving modes were then compared and analyzed. Drawing on the motion/force transmission index, we identified the good transmission workspace of the mechanism and performed a performance comparison analysis. The results unequivocally demonstrate that engaging the multi-drive mode substantially enhances the parallel mechanism's kinematic performance.
Gas Insulated Switchgear (GIS) equipment is an important substation device in the power system and plays an indispensable role in maintaining the operation of the power system. However, GIS equipment is prone to ablation defects and some foreign objects during long-term operation, thus affecting the normal operation of the power system. Aiming at the problems of low efficiency and easy-to-miss detection in manual inspection.This paper proposes a single-stage object detection model REP-YOLOX, which is based on structural re-parameterization and attention mechanisms technology. Firstly, we combine dilated convolution and standard convolution to extract features and obtain larger receptive fields. Then, in the Neck layer of the model, the Mask Conv block module is used to enhance the feature extraction ability under occlusion. Finally, the Transformer module is used in the prediction layer of the model to weigh the global features and predict the results through the full connection layer. Experimental results show that the model can achieve 99.14% mAP50 in the GIS equipment defect detection task, which indicates it can complete GIS equipment defect detection task well.
This paper proposes a dual-frequency image fusion algorithm to address the problem of poor preservation of image naturalness in GIS cavities. Firstly, the low-frequency image is obtained by low-pass filtering, and using an improved illumination estimating algorithm to improve low-frequency image brightness; At the same time, using homomorphic high-pass filtering suppresses the low-frequency components and enhances the high-frequency components of the image, and high-frequency image with increased details and uniform illumination is obtained; Then, linear fusion is used to perform linear fusion on the enhanced high and low-frequency images; Finally, experimental verification was conducted in GIS cavities environment, and the effectiveness of the image enhancement algorithm was verified by comparing multiple image evaluation indicators.
The human hand helps to perform most of our complex and fine movements because of its complex skeletal-muscular system and rich sensory. At the same time, the complex hand system is one of the reasons why rehabilitation of hand dysfunction after stroke is tricky. Virtual reality-based rehabilitation provides a new approach to hand function rehabilitation by mapping the patient’s movements into a rich virtual reality system and providing synergistic stimulation of visual, auditory and haptic senses to improve the patient’s motivation for rehabilitation, promote brain function reorganization and accelerate hand function rehabilitation. So the synergistic approach of multiple senses is very important. In this article, we designed a visually synergistic approach to flexible exoskeleton hand control with the aim of increasing the sense of active participation. We built a model of human-machine coupled motion, combined with a target grasping gesture obtained based on a virtual reality scene, and the target gesture is executed in the virtual system while controlling the synchronized motion of the exoskeleton hand. The results show that the motion of the virtual hand predicted by the human-computer coupling model is synchronized with the motion of the exoskeleton hand.
