Chengjin Zhang

Recent advances in the vision-language model (VLM) have demonstrated impressive capabilities in aligning high-level semantic information with visual environments, offering significant help for task planning in robotic manipulation. However, existing methods often fail to convert semantic tasks into precise robotic actions, especially in complex 3-D environments where keypoint localization and interaction dynamics are critical. To address this challenge, this article proposes principal component keypoint-reinforcement learning (PCK-RL), a multistage framework that bridges semantic reasoning and low-level control through semantic subplan decomposition, object keypoint localization, and example trajectory generation. Principal component analysis (PCA) is introduced to normalize the inputs and outputs of the VLM, constructing novel intermediate representations to enhance model performance. Additionally, a hybrid robotic action alignment method rapidly exploring random tree star-reinforcement learning (RRT*-RL) is introduced. It refines zero-shot reference trajectories by incorporating reinforcement learning (RL) with a closed-loop environmental feedback mechanism, which is crucial for long-horizon tasks. This article validates PCK-RL through extensive experiments in both simulated and real-world robotic environments, demonstrating its effectiveness and robustness in executing long-horizon tasks and manipulating diverse objects.

In recent years, intelligent fault diagnosis methods based on deep learning have developed rapidly. However, most of the existing work performs well under the assumption that training and testing samples are collected from the same distribution, and the performance drops sharply when the data distribution changes. For rolling bearings, the data distribution will change when the load and speed change. In this article, to improve fault diagnosis accuracy and anti-noise ability under different working loads, a transfer learning method based on multi-scale capsule attention network and joint distributed optimal transport (MSCAN-JDOT) is proposed for bearing fault diagnosis under different loads. Because multi-scale capsule attention networks can improve feature expression ability and anti-noise performance, the fault data can be better expressed. Using the domain adaptation ability of joint distribution optimal transport, the feature distribution of fault data under different loads is aligned, and domain-invariant features are learned. Through experiments that investigate bearings fault diagnosis under different loads, the effectiveness of MSCAN-JDOT is verified; the fault diagnosis accuracy is higher than that of other methods. In addition, fault diagnosis experiment is carried out in different noise environments to demonstrate MSCAN-JDOT, which achieves a better anti-noise ability than other transfer learning methods.

As critical components of DNA, enhancers can efficiently and specifically manipulate the spatial and temporal regulation of gene transcription. Malfunction or dysregulation of enhancers is implicated in a slew of human pathology. Therefore, identifying enhancers and their strength may provide insights into the molecular mechanisms of gene transcription and facilitate the discovery of candidate drug targets. In this paper, a new enhancer and its strength predictor, iEnhancer-GAN, is proposed based on a deep learning framework in combination with the word embedding and sequence generative adversarial net (Seq-GAN). Considering the relatively small training dataset, the Seq-GAN is designed to generate artificial sequences. Given that each functional element in DNA sequences is analogous to a “word” in linguistics, the word segmentation methods are proposed to divide DNA sequences into “words”, and the skip-gram model is employed to transform the “words” into digital vectors. In view of the powerful ability to extract high-level abstraction features, a convolutional neural network (CNN) architecture is constructed to perform the identification tasks, and the word vectors of DNA sequences are vertically concatenated to form the embedding matrices as the input of the CNN. Experimental results demonstrate the effectiveness of the Seq-GAN to expand the training dataset, the possibility of applying word segmentation methods to extract “words” from DNA sequences, the feasibility of implementing the skip-gram model to encode DNA sequences, and the powerful prediction ability of the CNN. Compared with other state-of-the-art methods on the training dataset and independent test dataset, the proposed method achieves a significantly improved overall performance. It is anticipated that the proposed method has a certain promotion effect on enhancer related fields.

The DNA replication influences the inheritance of genetic information in the DNA life cycle. As the distribution of replication origins (ORIs) is the major determinant to precisely regulate the replication process, the correct identification of ORIs is significant in giving an insightful understanding of DNA replication mechanisms and the regulatory mechanisms of genetic expressions. For eukaryotes in particular, multiple ORIs exist in each of their gene sequences to complete the replication in a reasonable period of time. To simplify the identification process of eukaryote’s ORIs, most of existing methods are developed by traditional machine learning algorithms, and target to the gene sequences with a fixed length. Consequently, the identification results are not satisfying, i.e. there is still great room for improvement. To break through the limitations in previous studies, this paper develops sequence segmentation methods, and employs the word embedding technique, ‘Word2vec’, to convert gene sequences into word vectors, thereby grasping the inner correlations of gene sequences with different lengths. Then, a deep learning framework to perform the ORI identification task is constructed by a convolutional neural network with an embedding layer. On the basis of the analysis of similarity reduction dimensionality diagram, Word2vec can effectively transform the inner relationship among words into numerical feature. For four species in this study, the best models are obtained with the overall accuracy of 0.975, 0.765, 0.885, 0.967, the Matthew’s correlation coefficient of 0.940, 0.530, 0.771, 0.934, and the AUC of 0.975, 0.800, 0.888, 0.981, which indicate that the proposed predictor has a stable ability and provide a high confidence coefficient to classify both of ORIs and non-ORIs. Compared with state-of-the-art methods, the proposed predictor can achieve ORI identification with significant improvement. It is therefore reasonable to anticipate that the proposed method will make a useful high throughput tool for genome analysis.

Deep learning based intelligent fault diagnosis methods have become a research hotspot in the fields of fault diagnosis and the health management of rolling bearings in recent years. To effectively identify incipient faults in rotating machinery, this paper proposes a novel hybrid intelligent fault diagnosis framework based on a convolutional neural network and support vector machine (SVM). First, an improved one-dimensional convolutional neural network (1DCNN) was adopted to extract fault features, and the state information and intrinsic properties of the raw vibration signals were mined. Second, the extracted features were used to train the SVM, which was applied to classify the fault category. The proposed hybrid framework combined the excellent classification performance of the SVM for small samples and the strong feature-learning ability of CNN network. In order to tune the parameters of the SVM, an improved novel particle swarm optimization algorithm (INPSO) which combined the Tent map and Lévy flight strategy was proposed. Numerical experimental results indicated that the proposed PSO variant had a better performance in searching accuracy and convergence speed. At last, multiple groups of rolling bearing fault diagnosis experiments were carried out and experimental results showed that, with the proposed 1DCNN-INPSO-SVM model, the hybrid framework was capable of diagnosing with high precision for rolling bearings and superior to some traditional fault diagnosis methods.

To solve the illumination sensitivity problems of mobile ground equipment, an enhanced visual SLAM algorithm based on the sparse direct method was proposed in this paper. Firstly, the vignette and response functions of the input sequences were optimized based on the photometric formation of the camera. Secondly, the Shi–Tomasi corners of the input sequence were tracked, and optimization equations were established using the pixel tracking of sparse direct visual odometry (VO). Thirdly, the Levenberg–Marquardt (L–M) method was applied to solve the joint optimization equation, and the photometric calibration parameters in the VO were updated to realize the real-time dynamic compensation of the exposure of the input sequences, which reduced the effects of the light variations on SLAM’s (simultaneous localization and mapping) accuracy and robustness. Finally, a Shi–Tomasi corner filtered strategy was designed to reduce the computational complexity of the proposed algorithm, and the loop closure detection was realized based on the oriented FAST and rotated BRIEF (ORB) features. The proposed algorithm was tested using TUM, KITTI, EuRoC, and an actual environment, and the experimental results show that the positioning and mapping performance of the proposed algorithm is promising.

Summary: Conventional simultaneous localization and mapping (SLAM) has concentrated on two-dimensional (2D) map building. To adapt it to urgent search and rescue (SAR) environments, it is necessary to combine the fast and simple global 2D SLAM and three-dimensional (3D) objects of interest (OOIs) local sub-maps. The main novelty of the present work is a method for 3D OOI reconstruction based on a 2D map, thereby retaining the fast performances of the latter. A theory is established that is adapted to a SAR environment, including the object identification, exploration area coverage (AC), and loop closure detection of revisited spots. Proposed for the first is image optical flow calculation with a 2D/3D fusion method and RGB-D (red, green, blue + depth) transformation based on Joblove–Greenberg mathematics and OpenCV processing. The mathematical theories of optical flow calculation and wavelet transformation are used for the first time to solve the robotic SAR SLAM problem. The present contributions indicate two aspects: (i) mobile robots depend on planar distance estimation to build 2D maps quickly and to provide SAR exploration AC; (ii) 3D OOIs are reconstructed using the proposed innovative methods of RGB-D iterative closest points (RGB-ICPs) and 2D/3D principle of wavelet transformation. Different mobile robots are used to conduct indoor and outdoor SAR SLAM. Both the SLAM and the SAR OOIs detection are implemented by simulations and ground-truth experiments, which provide strong evidence for the proposed 2D/3D reconstruction SAR SLAM approaches adapted to post-disaster environments.

An environment can be searched far more efficiently if the appropriate search strategy is used. Because of the limited individual abilities of swarm robots, namely, local sensing and low processing power, random searching is the main search strategy used in swarm robotics. The random walk methods that are used most commonly are Brownian motion and Lévy flight, both of which mimic the self-organized behavior of social insects. However, both methods are somewhat limited when applied to swarm robotics, where having the robots search repeatedly can result in highly inefficient searching. Therefore, by analyzing the characteristics of swarm robotic exploration, this paper proposes an improved random walk method in which each robot adjusts its step size adaptively to reduce the number of repeated searches by estimating the density of robots in the environment. Simulation experiments and experiments with actual robots are conducted to study the effectiveness of the proposed method and evaluate its performance in an exploration mission. The experimental results presented in this paper show that an area is covered more efficiently using the proposed method than it is using either Brownian motion or Lévy flight.