The growing adoption of deep learning, particularly supervised learning, in the manufacturing highlights the need for large labeled datasets. However, generating domain-specific labeled data is costly. Focusing on smart tightening diagnosis in manufacturing, prior research introduced the tightening diagnosis transformer (TDT), which leverages self-supervised transformers to reduce dependency on labeled data. Despite its advancements, TDT has two key limitations: (1) reliance solely on torque sensor data, ignoring angle sensor data, and (2) added computational overhead from self-supervised learning, which is problematic in resource-limited shop-floor environments. This study presents a novel transformer-based multi-label classification method that integrates sensor fusion and reduces needs for both computation and labeled data. We enhance the state-of-the-art TDT by introducing the angle positional encoder (APE), enabling feature-level sensor fusion for supervised learning. Additionally, we propose a self-supervised learning method for APE-enhanced TDT to reduce the need for extensive labeled datasets. We also introduce the random sequence patchifier (RSP), a transformer-specific data augmentation technique that improves generalization and reduces computational cost. Finally, we adopt annealing augmentation scheduling to mitigate the risk of learning “fake” feature representations (unrealistic artifacts created by the augmentations). Compared with previous TDT, our experiment evaluation demonstrates that the these introduced techniques improve Subset Accuracy and F1 scores by 10% and 7%. Moreover, the RSP-based augmentation reduces the training time by 12% for supervised learning and 15% for self-supervised learning.

This study focuses on the locomotion capability improvement in a tendon-driven soft quadruped robot through an online adaptive learning approach. Leveraging the inverse kinematics model of the soft quadruped robot, we employ a central pattern generator to design a parametric gait pattern, and use Bayesian optimization (BO) to find the optimal parameters. Further, to address the challenges of modeling discrepancies, we implement a multi-fidelity BO approach, combining data from both simulation and physical experiments throughout training and optimization. This strategy enables the adaptive refinement of the gait pattern and ensures a smooth transition from simulation to real-world deployment for the controller. Compared to previous result using a fixed gait pattern, the multi-fidelity BO approach improves the robot’s average walking speed from 0.14 m/s to 0.214 m/s, an increase of 52.7%. Moreover, we integrate a computational task off-loading architecture by edge computing, which reduces the onboard computational and memory overhead, to improve real-time control performance and facilitate an effective online learning process. The proposed approach successfully achieves optimal walking gait design for physical deployment with high efficiency, effectively addressing challenges related to the reality gap in soft robotics.

In recent years, enormous investments in Automated Driving Systems (ADSs) have distinctly advanced ADS technologies. Despite promises made by several high profile auto-makers, it has however become clear that the challenges involved for deploying ADS have been drastically underestimated. Contrary to previous generations of automotive systems, common design, development, verification and validation methods for safety critical systems do not suffice to cope with the increased complexity and operational uncertainties of an ADS. Therefore, the aim of this paper is to provide an understanding of existing methods for providing safety evidence and, most importantly, identifying the associated challenges and gaps pertaining to the use of each method. To this end, we have performed a literature review, articulated around four categories of methods: design techniques, verification and validation methods, run-time risk assessment, and run-time (self-)adaptation. We have identified and present eight challenges, collectively distinguishing ADSs from safety critical systems in general, and discuss the reviewed methods in the light of these eight challenges. For all reviewed methods, the uncertainties of the operational environment and the allocation of responsibility for the driving task on the ADS stand-out as the most difficult challenges to address. Finally, a set of research gaps is identified, and grouped into five major themes: (i) completeness of provided safety evidence, (ii) improvements and analysis needs, (iii) safe collection of closed loop data and accounting for tactical responsibility on the part of the ADS, (iv) integration of AI/ML-based components, and (v) scalability of the approaches with respect to the complexity of the ADS.

Introducing connectivity and collaboration promises to address some of the safety challenges for automated vehicles (AVs), especially in scenarios where occlusions and rule-violating road users pose safety risks and challenges in reconciling performance and safety. This requires establishing new collaborative systems with connected vehicles, off-board perception systems, and a communication network. However, adding connectivity and information sharing not only requires infrastructure investments but also an improved understanding of the design space, the involved trade-offs and new failure modes. We set out to improve the understanding of the relationships between the constituents of a collaborative system to investigate design parameters influencing safety properties and their performance trade-offs. To this end we propose a methodology comprising models, analysis methods, and a software tool for design space exploration regarding the potential for safety enhancements and requirements on off-board perception systems, the communication network, and AV tactical safety behavior. The methodology is instantiated as a concrete set of models and a tool, exercised through a case study involving intersection traffic conflicts. We show how the age of information and observation uncertainty affect the collaborative system design space and further discuss the generalization and other findings from both the methodology and case study development.

Autonomous agents rely on sensor data to construct representations of their environments, essential for predicting future events and planning their actions. However, sensor measurements suffer from limited range, occlusions, and sensor noise. These challenges become more evident in highly dynamic environments. This work proposes a probabilistic framework to jointly infer which parts of an environment are statically and which parts are dynamically occupied. We formulate the problem as a Bayesian network and introduce minimal assumptions that significantly reduce the complexity of the problem. Based on those, we derive Transitional Grid Maps (TGMs), an efficient analytical solution. Using real data, we demonstrate how this approach produces better maps than the state-of-the-art by keeping track of both static and dynamic elements and, as a side effect, can help improve existing SLAM algorithms.

This study provides an evaluation of an architecture framework intended to support stakeholders in realizing trustworthy cyber-physical systems (CPS), referred to as the T-Framework. The framework explicitly addresses CPS complexity, including the fact that multiple trustworthiness aspects will need to be considered for contemporary CPS, from classical dependability aspects to ethical concerns involving artificial intelligence. In addition, this study also investigates the problems that are repeatedly encountered by the stakeholders involved in realizing trustworthy CPS. To achieve the goals of the study, the boundary object and knowledge boundary concepts from social sciences were used. These concepts are useful tools to examine how various involved stakeholders can cooperate on a project through the utilization of objects, even though they have different perspectives and conflicting interests. Focus groups were used as the methodological approach to gather feedback from various experts in CPS from industry and academia. Findings show that stakeholders repeatedly encounter problems when making trade-offs between trustworthiness attributes and system aspects, dealing with prioritization, and making final decisions. The findings further show that the T-Framework can potentially guide stakeholders in addressing these problems as a boundary object. Furthermore, based on the feedback from the participants, several aspects for improvements or additional consideration in the T-Framework were identified, including clarifications regarding the framework workflow and terminology.

The state-of-the-art logic elements, like CPUs and FPGAs, are either trending towards higher core counts or heterogeneous architecture, driven primarily by the telecommunication and the automotive industry. The railway industry market, although relatively small, adopts logic elements from these leading sectors. However, there are significant challenges in adopting logic elements from other domains for the railway signaling systems which are expected to operate continuously for nearly thirty years ensuring the highest levels of safety. Furthermore, the intricacies involved in demonstrating safety prevent signaling applications from fully harnessing the capabilities of multi-core or heterogeneous architectures. Our paper focuses on the utilization constraints and challenges of using the latest logic elements in signaling applications conforming to EN 50129:2018. We analyze the use of such logic elements and architectural patterns to meet safety requirements cost-efficiently.

Soft actuators have the advantages of compliance and adaptability when working with vulnerable objects, but the deformation shape of the soft actuators is difficult to measure or estimate. Soft sensors made of highly flexible and responsive materials are promising new approaches to the shape estimation of soft actuators, but suffer from highly nonlinear, hysteresis, and time-variant properties. A nonlinear and adaptive state observer is essential for the shape estimation from soft sensors. Current state estimation methods rely on complex nonlinear data-fitting models, and the robustness of the estimation methods is questionable. This study investigates the soft actuator dynamics and the soft sensor model as a stochastic process characterized by the Gaussian Process (GP) model. The unscented Kalman filter (UKF) is applied to the GP model for more reliable variance adjustment during the sequential state estimation process than conventional methods. In addition, a major limitation of the GP model is its computational complexity during online inference. To improve the real-time performance while guaranteeing accuracy, we introduce an edge server to decrease the onboard computational and memory overhead. The experiments showcase a significant improvement in estimation accuracy and real-time performance compared to baseline methods.

Introducing connectivity and collaboration promises to address some of the safety challenges facing automated vehicles. This is especially the case for scenarios that suffer from the so-called information gap, where occlusions and rule violating road users pose safety risks, and challenges in reconciling performance and safety. 

Establishing new collaborative systems, encompassing connected vehicles, off-board perception systems and a communication network, can help to overcome the information gap, thus contributing to enhance traffic safety and performance. However, adding connectivity and information sharing not only requires infrastructure investments but also an improved understanding of the design space, the involved trade-offs and new failure modes. 

We set out to improve the understanding of the relationships between the constituents of a collaborative system to investigate design parameters influencing safety properties and their performance trade-offs. To this end we propose a methodology comprising models, analysis methods and a software tool that enables to explore the design space, the potential for safety enhancements, the corresponding requirements on off-board perception systems and the communication network, and tactical safety behaviors of connected automated vehicles. The methodology is instantiated in terms of a concrete set of models and a tool, exercised through a case study involving intersection traffic conflicts.    

We show how the age of information and measurement uncertainty affect the collaborative system design space and further discuss the generalization and other findings from both the methodology and case study development.