RESUMO
With the development of artificial intelligence, robots are widely used in various fields, grasping detection has been the focus of intelligent robot research. A dual manipulator grasping detection model based on Markov decision process is proposed to realize the stable grasping with complex multiple objects in this paper. Based on the principle of Markov decision process, the cross entropy convolutional neural network and full convolutional neural network are used to parameterize the grasping detection model of dual manipulators which are two-finger manipulator and vacuum sucker manipulator for multi-objective unknown objects. The data set generated in the simulated environment is used to train the two grasping detection networks. By comparing the grasping quality of the detection network output the best grasping by the two grasping methods, the network with better detection effect corresponding to the two grasping methods of two-finger and vacuum sucker is determined, and the dual manipulator grasping detection model is constructed in this paper. Robot grasping experiments are carried out, and the experimental results show that the proposed dual manipulator grasping detection method achieves 90.6% success rate, which is much higher than the other groups of experiments. The feasibility and superiority of the dual manipulator grasping detection method based on Markov decision process are verified.
Assuntos
Inteligência Artificial , Redes Neurais de Computação , Dedos , Extremidade Superior , Força da MãoRESUMO
sEMG(surface electromyography) signals have been widely used in rehabilitation medicine in the past decades because of their non-invasive, convenient and informative features, especially in human action recognition, which has developed rapidly. However, the research on sparse EMG in multi-view fusion has made less progress compared to high-density EMG signals, and for the problem of how to enrich sparse EMG feature information, a method that can effectively reduce the information loss of feature signals in the channel dimension is needed. In this paper, a novel IMSE (Inception-MaxPooling-Squeeze- Excitation) network module is proposed to reduce the loss of feature information during deep learning. Then, multiple feature encoders are constructed to enrich the information of sparse sEMG feature maps based on the multi-core parallel processing method in multi-view fusion networks, while SwT (Swin Transformer) is used as the classification backbone network. By comparing the feature fusion effects of different decision layers of the multi-view fusion network, it is experimentally obtained that the fusion of decision layers can better improve the classification performance of the network. In NinaPro DB1, the proposed network achieves 93.96% average accuracy in gesture action classification with the feature maps obtained in 300ms time window, and the maximum variation range of action recognition rate of individuals is less than 11.2%. The results show that the proposed framework of multi-view learning plays a good role in reducing individuality differences and augmenting channel feature information, which provides a certain reference for non-dense biosignal pattern recognition.
RESUMO
Simultaneous localization and mapping (SLAM) technology can be used to locate and build maps in unknown environments, but the constructed maps often suffer from poor readability and interactivity, and the primary and secondary information in the map cannot be accurately grasped. For intelligent robots to interact in meaningful ways with their environment, they must understand both the geometric and semantic properties of the scene surrounding them. Our proposed method can not only reduce the absolute positional errors (APE) and improve the positioning performance of the system but also construct the object-oriented dense semantic point cloud map and output point cloud model of each object to reconstruct each object in the indoor scene. In fact, eight categories of objects are used for detection and semantic mapping using coco weights in our experiments, and most objects in the actual scene can be reconstructed in theory. Experiments show that the number of points in the point cloud is significantly reduced. The average positioning error of the eight categories of objects in Technical University of Munich (TUM) datasets is very small. The absolute positional error of the camera is also reduced with the introduction of semantic constraints, and the positioning performance of the system is improved. At the same time, our algorithm can segment the point cloud model of objects in the environment with high accuracy.
Assuntos
Aprendizado Profundo , Imageamento Tridimensional , Algoritmos , Imageamento Tridimensional/métodosRESUMO
The continuous development of deep learning improves target detection technology day by day. The current research focuses on improving the accuracy of target detection technology, resulting in the target detection model being too large. The number of parameters and detection speed of the target detection model are very important for the practical application of target detection technology in embedded systems. This article proposed a real-time target detection method based on a lightweight convolutional neural network to reduce the number of model parameters and improve the detection speed. In this article, the depthwise separable residual module is constructed by combining depthwise separable convolution and non-bottleneck-free residual module, and the depthwise separable residual module and depthwise separable convolution structure are used to replace the VGG backbone network in the SSD network for feature extraction of the target detection model to reduce parameter quantity and improve detection speed. At the same time, the convolution kernels of 1 × 3 and 3 × 1 are used to replace the standard convolution of 3 × 3 by adding the convolution kernels of 1 × 3 and 3 × 1, respectively, to obtain multiple detection feature graphs corresponding to SSD, and the real-time target detection model based on a lightweight convolutional neural network is established by integrating the information of multiple detection feature graphs. This article used the self-built target detection dataset in complex scenes for comparative experiments; the experimental results verify the effectiveness and superiority of the proposed method. The model is tested on video to verify the real-time performance of the model, and the model is deployed on the Android platform to verify the scalability of the model.
RESUMO
The analysis of robot inverse kinematic solutions is the basis of robot control and path planning, and is of great importance for research. Due to the limitations of the analytical and geometric methods, intelligent algorithms are more advantageous because they can obtain approximate solutions directly from the robot's positive kinematic equations, saving a large number of computational steps. Particle Swarm Algorithm (PSO), as one of the intelligent algorithms, is widely used due to its simple principle and excellent performance. In this paper, we propose an improved particle swarm algorithm for robot inverse kinematics solving. Since the setting of weights affects the global and local search ability of the algorithm, this paper proposes an adaptive weight adjustment strategy for improving the search ability. Considering the running time of the algorithm, this paper proposes a condition setting based on the limit joints, and introduces the position coefficient k in the velocity factor. Meanwhile, an exponential product form modeling method (POE) based on spinor theory is chosen. Compared with the traditional DH modeling method, the spinor approach describes the motion of a rigid body as a whole and avoids the singularities that arise when described by a local coordinate system. In order to illustrate the advantages of the algorithm in terms of accuracy, time, convergence and adaptability, three experiments were conducted with a general six-degree-of-freedom industrial robotic arm, a PUMA560 robotic arm and a seven-degree-of-freedom robotic arm as the research objects. In all three experiments, the parameters of the robot arm, the range of joint angles, and the initial attitude and position of the end-effector of the robot arm are given, and the attitude and position of the impact point of the end-effector are set to verify whether the joint angles found by the algorithm can reach the specified positions. In Experiments 2 and 3, the algorithm proposed in this paper is compared with the traditional particle swarm algorithm (PSO) and quantum particle swarm algorithm (QPSO) in terms of position and direction solving accuracy, operation time, and algorithm convergence. The results show that compared with the other two algorithms, the algorithm proposed in this paper can ensure higher position accuracy and orientation accuracy of the robotic arm end-effector. the position error of the algorithm proposed in this paper is 0 and the maximum orientation error is 1.29 × 10-8. while the minimum position error of the other two algorithms is -1.64 × 10-5 and the minimum orientation error is -4.03 × 10-6. In terms of operation time, the proposed algorithm in this paper has shorter operation time compared with the other two algorithms. In the last two experiments, the computing time of the proposed algorithm is 0.31851 and 0.30004s respectively, while the shortest computing time of the other two algorithms is 0.33359 and 0.30521s respectively. In terms of algorithm convergence, the proposed algorithm can achieve faster and more stable convergence than the other two algorithms. After changing the experimental subjects, the proposed algorithm still maintains its advantages in terms of accuracy, time and convergence, which indicates that the proposed algorithm is more applicable and has certain potential in solving the multi-arm inverse kinematics solution. This paper provides a new way of thinking for solving the multi-arm inverse kinematics solution problem.
RESUMO
The development of object detection technology makes it possible for robots to interact with people and the environment, but the changeable application scenarios make the detection accuracy of small and medium objects in the practical application of object detection technology low. In this paper, based on multi-scale feature fusion of indoor small target detection method, using the device to collect different indoor images with angle, light, and shade conditions, and use the image enhancement technology to set up and amplify a date set, with indoor scenarios and the SSD algorithm in target detection layer and its adjacent features fusion. The Faster R-CNN, YOLOv5, SSD, and SSD target detection models based on multi-scale feature fusion were trained on an indoor scene data set based on transfer learning. The experimental results show that multi-scale feature fusion can improve the detection accuracy of all kinds of objects, especially for objects with a relatively small scale. In addition, although the detection speed of the improved SSD algorithm decreases, it is faster than the Faster R-CNN, which better achieves the balance between target detection accuracy and speed.
RESUMO
In order to solve the problems of poor image quality, loss of detail information and excessive brightness enhancement during image enhancement in low light environment, we propose a low-light image enhancement algorithm based on improved multi-scale Retinex and Artificial Bee Colony (ABC) algorithm optimization in this paper. First of all, the algorithm makes two copies of the original image, afterwards, the irradiation component of the original image is obtained by used the structure extraction from texture via relative total variation for the first image, and combines it with the multi-scale Retinex algorithm to obtain the reflection component of the original image, which are simultaneously enhanced using histogram equalization, bilateral gamma function correction and bilateral filtering. In the next part, the second image is enhanced by histogram equalization and edge-preserving with Weighted Guided Image Filtering (WGIF). Finally, the weight-optimized image fusion is performed by ABC algorithm. The mean values of Information Entropy (IE), Average Gradient (AG) and Standard Deviation (SD) of the enhanced images are respectively 7.7878, 7.5560 and 67.0154, and the improvement compared to original image is respectively 2.4916, 5.8599 and 52.7553. The results of experiment show that the algorithm proposed in this paper improves the light loss problem in the image enhancement process, enhances the image sharpness, highlights the image details, restores the color of the image, and also reduces image noise with good edge preservation which enables a better visual perception of the image.
RESUMO
Complete trajectory planning includes path planning, inverse solution solving and trajectory optimization. In this paper, a highly smooth and time-saving approach to trajectory planning is obtained by improving the kinematic and optimization algorithms for the time-optimal trajectory planning problem. By partitioning the joint space, the paper obtains an inverse solution calculation based on the partitioning of the joint space, saving 40% of the inverse kinematics solution time. This means that a large number of computational resources can be saved in trajectory planning. In addition, an improved sparrow search algorithm (SSA) is proposed to complete the solution of the time-optimal trajectory. A Tent chaotic mapping was used to optimize the way of generating initial populations. The algorithm was further improved by combining it with an adaptive step factor. The experiments demonstrated the performance of the improved SSA. The robot's trajectory is further optimized in time by an improved sparrow search algorithm. Experimental results show that the method can improve convergence speed and global search capability and ensure smooth trajectories.
RESUMO
Autonomous Underwater Vehicle are widely used in industries, such as marine resource exploitation and fish farming, but they are often subject to a large amount of interference which cause poor control stability, while performing their tasks. A decoupling control algorithm is proposed and A single control volume-single attitude angle model is constructed for the problem of severe coupling in the control system of attitude of six degrees of freedom Autonomous Underwater Vehicle. Aiming at the problem of complex Active Disturbance Rejection Control (ADRC) adjustment relying on manual experience, the PSO-ADRC algorithm is proposed to realize the automatic adjustment of its parameters, which improves the anti-interference ability and control accuracy of Autonomous Underwater Vehicle in dynamic environment. The anti-interference ability and control accuracy of the method were verified through experiments.
RESUMO
Gesture recognition technology is widely used in the flexible and precise control of manipulators in the assisted medical field. Our MResLSTM algorithm can effectively perform dynamic gesture recognition. The result of surface EMG signal decoding is applied to the controller, which can improve the fluency of artificial hand control. Much current gesture recognition research using sEMG has focused on static gestures. In addition, the accuracy of recognition depends on the extraction and selection of features. However, Static gesture research cannot meet the requirements of natural human-computer interaction and dexterous control of manipulators. Therefore, a multi-stream residual network (MResLSTM) is proposed for dynamic hand movement recognition. This study aims to improve the accuracy and stability of dynamic gesture recognition. Simultaneously, it can also advance the research on the smooth control of the Manipulator. We combine the residual model and the convolutional short-term memory model into a unified framework. The architecture extracts spatiotemporal features from two aspects: global and deep, and combines feature fusion to retain essential information. The strategy of pointwise group convolution and channel shuffle is used to reduce the number of network calculations. A dataset is constructed containing six dynamic gestures for model training. The experimental results show that on the same recognition model, the gesture recognition effect of fusion of sEMG signal and acceleration signal is better than that of only using sEMG signal. The proposed approach obtains competitive performance on our dataset with the recognition accuracies of 93.52%, achieving state-of-the-art performance with 89.65% precision on the Ninapro DB1 dataset. Our bionic calculation method is applied to the controller, which can realize the continuity of human-computer interaction and the flexibility of manipulator control.
RESUMO
Gesture recognition is one of the important ways of human-computer interaction, which is mainly detected by visual technology. The temporal and spatial features are extracted by convolution of the video containing gesture. However, compared with the convolution calculation of a single image, multiframe image of dynamic gestures has more computation, more complex feature extraction, and more network parameters, which affects the recognition efficiency and real-time performance of the model. To solve above problems, a dynamic gesture recognition model based on CBAM-C3D is proposed. Key frame extraction technology, multimodal joint training, and network optimization with BN layer are used for making the network performance better. The experiments show that the recognition accuracy of the proposed 3D convolutional neural network combined with attention mechanism reaches 72.4% on EgoGesture dataset, which is improved greatly compared with the current main dynamic gesture recognition methods, and the effectiveness of the proposed algorithm is verified.
Assuntos
Gestos , Reconhecimento Automatizado de Padrão , Algoritmos , Humanos , Redes Neurais de Computação , Reconhecimento PsicológicoRESUMO
With the manipulator performs fixed-point tasks, it becomes adversely affected by external disturbances, parameter variations, and random noise. Therefore, it is essential to improve the robust and accuracy of the controller. In this article, a self-tuning particle swarm optimization (PSO) fuzzy PID positioning controller is designed based on fuzzy PID control. The quantization and scaling factors in the fuzzy PID algorithm are optimized by PSO in order to achieve high robustness and high accuracy of the manipulator. First of all, a mathematical model of the manipulator is developed, and the manipulator positioning controller is designed. A PD control strategy with compensation for gravity is used for the positioning control system. Then, the PID controller parameters dynamically are minute-tuned by the fuzzy controller 1. Through a closed-loop control loop to adjust the magnitude of the quantization factors-proportionality factors online. Correction values are outputted by the modified fuzzy controller 2. A quantization factor-proportion factor online self-tuning strategy is achieved to find the optimal parameters for the controller. Finally, the control performance of the improved controller is verified by the simulation environment. The results show that the transient response speed, tracking accuracy, and follower characteristics of the system are significantly improved.