Bearings frequently operate under time-varying working conditions where speed profiles may be unseen for intelligent diagnostic models. Frontier research succeeds in extracting discriminative and invariant features for accurate diagnosis. However, their extraction and generalization mechanisms lack physical interpretations, resulting in dubious generalizability and high data dependency. To tackle the challenges in practical scenarios, this study incorporates fault mechanism, proposing a fully interpretable network for credible fault diagnosis across unseen time-varying working conditions, called convolutional sparse modal unrolling network (CSMUNet). For interpretable time-varying feature extraction, a novel sparse coding algorithm is conceived and forms CSMUNet via algorithm unrolling. The algorithm utilizes modal response as convolutional dictionary for sparse vector optimization, under a masking constraint of time-varying impulsive moments. To generalize across unseen working conditions, an interpretable domain-invariant representation is conceived based on impulsive fault mechanism. The input is divided by equiangular span instead of time length, enabling impulses of different samples to have constant numbers for accurate moment recognition in CSMUNet. The proposed model is tested under unseen time-varying speed with simulation and experiment data. Results demonstrate the superior discriminative feature extraction, with key metrics rising to 3.2 times higher than raw inputs and classical convolutional models reaching over 35% accuracy improvement. The performance of CSMUNet is interpreted with the domain-invariant representation, which enhances its diagnostic credibility.

Multi-robot global localization (MR-GL) with unknown initial positions in a large scale environment is a challenging task. The key point is the data association between different robots’ viewpoints. It also makes traditional Appearance-based localization methods unusable. Recently, researchers have utilized the object’s semantic invariance to generate a semantic graph to address this issue. However, previous works lack robustness and are sensitive to overlap rate of maps, resulting in unpredictable performance in real-world environments. In this paper, we propose a data association algorithm based on neighbor constraints to improve the robustness of the system. We demonstrate the effectiveness of our method on three different datasets, indicating a significant improvement in robustness compared to previous works.

Motor skills are essential for robots to accomplish complicated and dexterous manipulation tasks, which are difficult to be mastered through traditional controller designs. Currently, robots learning from demonstrations enable them to learn control policies automatically from human motor demonstrations. However, the nonlinearity and instantaneousness of the demonstrated forces prohibit robots from fully mastering the motor skill features by simply exploiting force examples. Therefore, a self-improvement learning scheme is required to refine the control policy further until satisfactory motor skills are acquired. Hence, this paper combines learning from demonstrations and reinforcement learning to learn a controller for complex motor skills. The proposed method is validated on an IIWA KUKA robot, performing a specified nonprehensile manipulation task.Note to Practitioners—The motivation of this paper originates from the requirement to develop an efficient and fast learning algorithm that improves the robot skill learning efficiency. Specifically, our research focuses on the nonprehensile manipulation task, easily subject to environmental changes. Therefore, the robot must continuously interact with the environment to master the skill. To accelerate the skill learning process, learning from demonstrations initializes the control policies, and then the robot starts to practice the demonstrated skill. After each practice round, the robot receives a reward from the environment, and based on the reinforcement learning algorithm limited up to 100 trials, the robot masters the nonprehensile manipulation skill.

6D pose estimation for objects is an important technology in human-computer interaction. Previous works trained one or more complicated networks to predict 6D poses. Although complex models have nice performance generally, the high storage and computation cost make it difficult to be applied on hardware platforms with limited computing ability such as the low-cost mobile terminal. Hence, how to reduce the complexity of the model while maintaining accuracy remains a challenge. In this paper, we present a lightweight generic architect that processes the color and depth images respectively by employing two efficient backbone networks, then use a fusion network to realize pose regression. Furthermore, an iterative refinement network compressed is implemented by using the Filter Pruning via Geometric Median (FPGM) algorithm to refine the poses while improving real-time performance. Comprehensive experiments conducted on two benchmark datasets, LineMOD and YCB-Video, confirm that the proposed model is more than twice as fast as the state-of-the-art (SOTA) DenseFusion. For main metrics, the BFLOPs (Billion FLoat OPerations) are reduced by 97.0%, and the parameter size declines by 87.4%. The average distance (ADD) for LineMOD increases by 2.6%. The overall performance of the new model is proven outperforming SOTA methods both in efficiency and accuracy.

Flexible job shop scheduling problem (FJSP) has long been a complex problem due to the resource flexibility and strong constraints, generating a mixed-integer non-linear optimization problem. The problem becomes more complex with the increasing demand of energy reduction and the corresponding environmental impacts. Proper production scheduling is of significant potential in saving energy in the manufacturing system. In this paper, a multi-objective FJSP model is formulated with the objectives of minimizing the makespan and energy consumption considering strong transportation constraints. Two popular multi-objective optimization solver including Non-dominated Sorting Genetic Algorithm-II (NSGA-II) and A Multiobjective Evolutionary Algorithm Based on Decomposition (MOEA/D) are employed and compared in a real-world instance of the FJSP, associated with novel coding schemes. The results show that the proposed model is well solved by the two solvers and NSGA-II get the better solutions.

The advent of myoelectric control schemes provides promising chances for locomotion empowerment and restoration of those with disabilities. Despite substantial efforts have been made into advancing sEMG-based motion recognition, it may be a little tricky to determine appropriate muscles and features for people with muscle disorders or different muscle use preferences. To mitigate it, an advantageous sEMG channel and feature selection method based on ReliefF algorithm was proposed. Related experiments were conducted on a eight able-bodied subject database to showcase the feasibility and efficiency of the proposed approach, that is, considerably high classification performance was maintained with the original feature set reduced by more than half. Ulteriorly, we also investigated the influences of different number of neighbors or features on classification accuracy for ascertaining the optimal values. The strengths of our proposed method lie in not only customizing channel and feature selection for individual users, but also offering preliminary insight for a general mapping mechanism between human muscles and corresponding motions.

Learning from demonstration (LfD) has been increasingly used to encode robot tasks such that robots can achieve reproduction more flexibly in unstructured environments (e.g., households or factories). It is an effective alternative to preprogramming methods owing to its capacity of enabling robots to generalize to different situations. In this paper, we focus on LfD in the point-to-point movement case, where the dilemma of stability and accuracy exists. To avoid such a dilemma, we propose a learning approach that guarantees accuracy and stability simultaneously by means of constructed manifold immersion and submersion. We evaluate the proposed approach on two libraries of human handwriting motions (the LASA data set and a self-made GREEK data set) and on a set of experiments on the Barrett WAM robot.

Crowdsensing leverages human intelligence/experience from the general public and social interactions to create participatory sensor networks, where context-aware and semantically complex information is gathered, processed, and shared to collaboratively solve specific problems. This paper proposes a real-time projector-camera finger system based on the crowdsensing, in which user can interact with a computer by bare hand touching on arbitrary surfaces. The interaction process of the system can be completely carried out automatically, and it can be used as an intelligent device in intelligent transport system where the driver can watch and interact with the display information while driving, without causing visual distractions. A single camera is used in the system to recover 3D information of fingertip for hand touch detection. A linear-scanning method is used in the system to determine the touch for increasing the users’ collaboration and operationality. Experiments are performed to show the feasibility of the proposed system. The system is robust to different lighting conditions. The average percentage of correct hand touch detection of the system is 92.0% and the average time of processing one video frame is 30 milliseconds.

Most current approaches in action recognition face difficulties that cannot handle recognition of multiple actions, fusion of multiple features, and recognition of action in frame by frame model, incremental learning of new action samples and application of position information of space-time interest points to improve performance simultaneously. In this paper, we propose a novel approach based on Position-Tree that takes advantage of the relationship of the position of joints and interest points. The normalized position of interest points indicates where the movement of body part has occurred. The extraction of local feature encodes the shape of the body part when performing action, justifying body movements. Additionally, we propose a new local descriptor calculating the local energy map from spatial-temporal cuboids around interest point. In our method, there are three steps to recognize an action: (1) extract the skeleton point and space-time interest point, calculating the normalized position according to their relationships with joint position, (2) extract the LEM (Local Energy Map) descriptor around interest point, (3) recognize these local features through non-parametric nearest neighbor and label an action by voting those local features. The proposed approach is tested on publicly available MSRAction3D dataset, demonstrating the advantages and the state-of-art performance of the proposed method.

Hand posture is a natural and effective human robot interaction way. In this paper, an user-independent hand posture recognition system using depth and color images captured from an RGB-D camera is presented. To recognize hand posture against complicated background conditions, we propose a novel method for automatic and accurate hand posture segmentation which detects the hand with Chamfer matching, tracks the hand with Kalman filter and segments the hand with region growing algorithm only in the depth space. A new hand posture descriptor invariant to scale, shift and in-plane rotation is constructed with the combination of local contour Fourier descriptor and global Bag-of-Features (BoF) descriptor based on Scale Invariance Feature Transform (SIFT). The sparse representation-based classification (SRC) is applied to perform the hand posture recognition task in the system. Experiments with a self-built large scale hand posture database collected online show the robustness and effectiveness of the proposed system.