Advanced intelligent systems (Weinheim an der Bergstrasse, Germany)最新文献

Imitation learning (IL) facilitates intuitive robotic programming. However, ensuring the reliability of learned behaviors remains a challenge. In the context of reaching motions, a robot should consistently reach its goal, regardless of its initial conditions. To meet this requirement, IL methods often employ specialized function approximators that guarantee this property by construction. Although effective, these approaches come with some limitations: 1) they are typically restricted in the range of motions they can model, resulting in suboptimal IL capabilities, and 2) they require explicit extensions to account for the geometry of motions that consider orientations. To address these challenges, we introduce a novel stability loss function that does not constrain the function approximator's architecture and enables learning policies that yield accurate results. Furthermore, it is not restricted to a specific state space geometry; therefore, it can easily incorporate the geometry of the robot's state space. Proof of the stability properties induced by this loss is provided and the method is empirically validated in various settings. These settings include Euclidean and non-Euclidean state spaces, as well as first-order and second-order motions, both in simulation and with real robots. More details about the experimental results can be found at https://youtu.be/ZWKLGntCI6w.

Head pose estimation plays a crucial role in various applications, including human–machine interaction, autonomous driving systems, and 3D reconstruction. Current methods address the problem primarily from a 2D perspective, which limits the efficient utilization of 3D information. Herein, a novel approach, called pose orientation-aware network (POANet), which leverages normal maps for orientation information embedding, providing abundant and robust head pose information, is introduced. POANet incorporates the axial signal perception module and the rotation matrix perception module, these lightweight modules make the approach achieve state-of-the-art (SOTA) performance with few computational costs. This method can directly analyze various topological 3D data without extensive preprocessing. For depth images, POANet outperforms existing methods on the Biwi Kinect head pose dataset, reducing the mean absolute error (MAE) by ≈30% compared to the SOTA methods. POANet is the first method to perform rigid head registration in a landmark-free manner. It also incorporates few-shot learning capabilities and achieves an MAE of about $��1�°�$ on the Headspace dataset. These features make POANet a superior alternative to traditional generalized Procrustes analysis for mesh data processing, offering enhanced convenience for human phenotype studies.

Pancreatic ductal adenocarcinoma (PDAC) poses a considerable diagnostic and therapeutic challenge due to the lack of specific biomarkers and late diagnosis. Early detection is crucial for improving prognosis, but current techniques are insufficient. An innovative approach based on differential scanning calorimetry (DSC) of blood serum samples, thermal liquid biopsy (TLB), combined with machine-learning (ML) analysis, may offer a more efficient method for diagnosing PDAC. Serum samples from a cohort of 212 PDAC patients and 184 healthy controls are studied. DSC thermograms are analyzed using ML models. The generated models are built applying algorithms based on penalized regression, resampling, categorization, cross validation, and variable selection. The ML-based model demonstrates outstanding ability to discriminate between PDAC patients and control subjects, with a sensitivity of 90% and an area under the ROC receiver operating characteristic curve of 0.83 in the training and test groups. Application of the model to an independent validation cohort of 113 PDAC patients confirms its robustness and utility as a diagnosis tool. The application of ML to serum TLB data emerges as a promising  methodology for early diagnosis, representing a significant advance for detecting and managing PDAC, envisaging a minimally invasive and more efficient methodology for identifying biomarkers.

This article presents a novel 3-axis Halbach permanent magnet elastomer (H-PME) sensor for the robotic application, which effectively reduces crosstalk along when two sensors are used simultaneously in close proximity, for example, during grasping of thin and delicate objects, needle threading, etc. This sensor integrates a Halbach-array magnetic elastomer, a 3×3 Hall sensor matrix, and a silicone layer. The magnetic elastomer is produced by combining NdFeB powders with a diameter of 5 μm into silicone, following a weight ratio of 50%, and then magnetized using 2D Halbach-array magnets. Simulation results reveal the capability to adjust magnetic field strength and distribution by altering the magnet's orientation. The sensor's efficacy in 3-axis sensing is validated through calibration with a linear model, achieving a good root-mean-square error below 0.7 N in force measurement. The H-PME sensor, with a thickness of merely 4.5 mm, can detect forces up to 50 N. It's simple 3-layer design allows the thickness to be reduced to as low as 2 mm, while also offering ease of replacement. Crucially, crosstalk evaluation experiments show that the proposed H-PME sensor can dramatically mitigate crosstalk interference.

Optical encryption is pivotal in information security, offering parallel processing, speed, and robust security. The simplicity and compatibility of speckle-based cryptosystems have garnered considerable attention. Yet, the predictable statistical distribution of speckle optical fields’ characteristics can invite statistical attacks, undermining these encryption methods. The proposed solution, a deep adversarial learning-based speckle modulation network (DeepSLM), disrupts the strong intercorrelation of speckle grains. Utilizing the unique encoding properties of speckle patterns, DeepSLM facilitates license editing within the modulation phase, pioneering a layered authentication encryption system. Our empirical studies confirm DeepSLM's superior performance on key metrics. Notably, the testing dataset reveals an average Pearson correlation coefficient above 0.97 between decrypted images and their original counterparts for intricate subjects like human faces, attesting to the method's high fidelity. This innovation marries adjustable modification, optical encryption, and deep learning to enforce tiered data access control, charting new paths for creating user-specific access protocols.

The rapid expansion of information systems in all areas of society demands more powerful, efficient, and low-energy consumption computing systems. Neuromorphic engineering has emerged as a solution that attempts to mimic the brain to incorporate its capabilities to solve complex problems in a computationally and energy-efficient way in real time. Within neuromorphic computing, building systems to efficiently store the information is still a challenge. Among all the brain regions, the hippocampus stands out as a short-term memory capable of learning and recalling large amounts of information quickly and efficiently. Herein, a spike-based bio-inspired hippocampus sequential memory model is proposed that makes use of the benefits of analog computing and spiking neural networks (SNNs): noise robustness, improved real-time operation, and energy efficiency. This model is applied to robotic navigation to learn and recall trajectories that lead to a goal position within a known grid environment. The model is implemented on the special-purpose SNNs mixed-signal DYNAP-SE hardware platform. Through extensive experimentation together with an extensive analysis of the model's behavior in the presence of external noise sources, its correct functioning is demonstrated, proving the robustness and consistency of the proposed neuromorphic sequential memory system.

Explainable artificial intelligence (XAI) aims to develop transparent explanatory approaches for “black-box” deep learning models. However, it remains difficult for existing methods to achieve the trade-off of the three key criteria in interpretability, namely, reliability, understandability, and usability, which hinder their practical applications. In this article, we propose a self-supervised automatic semantic interpretable explainable artificial intelligence (AS-XAI) framework, which utilizes transparent orthogonal embedding semantic extraction spaces and row-centered principal component analysis (PCA) for global semantic interpretation of model decisions in the absence of human interference, without additional computational costs. In addition, the invariance of filter feature high-rank decomposition is used to evaluate model sensitivity to different semantic concepts. Extensive experiments demonstrate that robust and orthogonal semantic spaces can be automatically extracted by AS-XAI, providing more effective global interpretability for convolutional neural networks (CNNs) and generating human-comprehensible explanations. The proposed approach offers broad fine-grained extensible practical applications, including shared semantic interpretation under out-of-distribution (OOD) categories, auxiliary explanations for species that are challenging to distinguish, and classification explanations from various perspectives. In a systematic evaluation by users with varying levels of AI knowledge, AS-XAI demonstrated superior “glass box” characteristics.

Multimodal communication is a human feature that enables diverse interactions. In human–robot interaction (HRI), robots have to communicate using human skills so that they can seem natural and assist effectively. Most research uses predefined gestures to equip robots with social abilities. However, researchers scarcely consider generating bioinspired involuntary behavior to improve a robot's expressiveness and communication. Human studies revealed that involuntary behavior affects how others perceive communicative intentions. Therefore, mimicking human involuntary behavior may positively affect HRI. This article extends our previous work on equipping robots with involuntary behavior with a user study that evaluates the use of bioinspiration for complementing gestures. A preliminary test is conducted with 15 participants to determine if they can perceive the intensities of the involuntary processes heart rate, pupil size, blink rate, breathing rate, and motor activity. 63 new participants interacted with a robot with bioinspired behaviors or a robot only showing predefined gestures to evaluate the robots’ warmth, competence, and discomfort. The results show that the preliminary test participants differentiated the intensities of the involuntary processes. Participants in the second study find the robot with bioinspired behaviors significantly warmer and more competent than the robot with predefined gestures, with no discomfort difference.

The advent of 3D printing has transformed manufacturing. However, extending the library of materials to improve 3D printing quality remains a challenge. Defects can occur when printing parameters like print speed and temperature are chosen incorrectly. These can cause structural or dimensional issues in the final product. This review investigates closed-loop artificial intelligence-augmented additive manufacturing (AI2AM) technology that integrates AI-based monitoring, automation, and optimization of printing parameters and processes. AI2AM uses AI to improve defect detection and prevention, improving additive manufacturing quality and efficiency. This article explores generic 3D printing processes and issues using existing research and developments. Next, it focuses on fused deposition modeling (FDM) printers and reviews their parameters and issues. The current remedies developed for defect detection and monitoring in FDM 3D printers are presented. Then, the article investigates AI-based 3D printing monitoring, closed-loop feedback systems, and parameter optimization development. Finally, closed-loop 3D printing challenges and future directions are discussed. AI-based systems detect and correct 3D printing failures, enabling current printers to operate within optimal conditions and minimizing the risk of defects or failures, which in turn leads to more sustainable manufacturing with minimum waste and extending the library of materials.

This article presents an extended Cartesian space robot control framework that features a virtual force tracking impedance control to enhance the end-effector trajectory tracking performance. Initially, the concept of a virtual surface is introduced, which is assumed to be at some constant distance from the desired end-effector trajectory. This virtual surface generates a virtual contact force when interacting with the torque-controlled robot end-effector. The interaction is then manipulated using an impedance control model to track a constant desired force. If the robot end-effector deviates from the desired trajectory, the constant force-tracking impedance control generates a compliance trajectory that regulates the end-effector movements, constraining it to the desired trajectory. For robust force tracking, impedance parameters are optimally tuned using a closed-loop dynamic model incorporating both robot and impedance dynamics. Additionally, super twisting sliding mode control (STSMC) is integrated to overcome uncertainties and the impact of robot dynamics on force-tracking performance. Experimental validation confirms the theoretical claims of the proposed approach. It demonstrates that force-tracking impedance control improves the end-effector trajectory tracking by quickly reacting to the dynamic trajectories compared to position control only and effectively maintains it on the desired trajectories.