Driver fatigue evaluation is of great importance for traffic safety and many intricate factors would exacerbate the difficulty. In this paper, based on the spatial–temporal structure of multichannel electroencephalogram (EEG) signals, we develop a novel EEG-based spatial–temporal convolutional neural network (ESTCNN) to detect driver fatigue. First, we introduce the core block to extract temporal dependencies from EEG signals. Then, we employ dense layers to fuse spatial features and realize classification. The developed network could automatically learn valid features from EEG signals, which outperforms the classical two-step machine learning algorithms. Importantly, we carry out fatigue driving experiments to collect EEG signals from eight subjects being alert and fatigue states. Using 2800 samples under within-subject splitting, we compare the effectiveness of ESTCNN with eight competitive methods. The results indicate that ESTCNN fulfills a better classification accuracy of 97.37% than these compared methods. Furthermore, the spatial–temporal structure of this framework advantages in computational efficiency and reference time, which allows further implementations in the brain–computer interface online systems.

In industrial production, current pipe-climbing robots can replace humans in real-time monitoring, leak inspection, and other tasks. However, most pipe-climbing robots are designed for specific applications with complex structures; this limits the adaptability to various conditions. To solve this issue, this letter presents a one-piece 3D-printed pipe-climbing robot composed of sequenced novel soft bending mechanisms. The mechanisms use the small strains of soft materials to achieve large bending deformation. The proposed robot is highly adaptable to pipes with various diameters, radii of curvature, inclinations, and even outer/inner walls because of modular grippers and an inchworm inspired middle section. The good strength and flexibility of the robot are simultaneously achieved by reasonable structural design using one material. In this letter, the design concept and processing methods are introduced. The finite element method is adopted to predict the maximum bending angle of modules to aid design, and a torque test bench is built to acquire output torque and restoring torque. Then, the model of the entire robot is built to facilitate specialized design. Finally, several experiments are conducted to verify the good adaptability of the robot to pipes of various diameters (18–36 mm), inclinations (0–90°, even horizontal pipes), and radii of curvature (minimum to 19 mm). The static bearing ability of the robot reaches 1000g (without climbing at 7 bar, 22 mm diameter vertical pipes), which is almost 80 times its weight. In addition, the load-bearing ability of the robot while climbing on pipes in various states is obtained.

Retinal microsurgery requires precise manipulation of delicate tissue in the interior of the eye. Force sensing instruments can provide surgeons the imperceptible force information, thereby improving the safety of retinal microsurgery. Previous force sensing microneedles have demonstrated robust performance in transverse force measurement. However, in some surgical operations such as puncturing the vessel, the axial force sensing function is critical. In this article, we present a three degrees of freedom force sensing microneedle based on fiber Bragg grating (FBG) sensors. A novel configuration of the FBG sensors is arranged to detect the transverse and axial forces. Three optical fibers with dual FBGs are longitudinally attached along the outside surface of a Nitinol tube. Benefiting from the novel configuration of the FBG sensors, a hollow channel can be used to pass through injection catheter. A new force calculation algorithm was developed to decouple the transverse forces and axial forces. Experimental results demonstrate the proposed force calculation algorithm can achieve a resolution of 0.124 and 0.74 mN for transverse forces and axial forces, respectively. Validation and temperature compensation experiments show the force sensor can provide consistent and accurate measurement of 3-D forces within 35–37 °C temperature variation.

Purpose: Upper gastrointestinal (GI) endoscopic image documentation has provided an efficient, low-cost solution to address quality control for endoscopic reporting. The problem is, however, challenging for computer-assisted techniques, because different sites have similar appearances. Additionally, across different patients, site appearance variation may be large and inconsistent. Therefore, according to the British and modified Japanese guidelines, we propose a set of oesophagogastroduodenoscopy (EGD) images to be routinely captured and evaluate its efficiency for deep learning-based classification methods. Methods: A novel EGD image dataset standardising upper GI endoscopy to several steps is established following landmarks proposed in guidelines and annotated by an expert clinician. To demonstrate the discrimination of proposed landmarks that enable the generation of an automated endoscopic report, we train several deep learning-based classification models utilising the well-annotated images. Results: We report results for a clinical dataset composed of 211 patients (comprising a total of 3704 EGD images) acquired during routine upper GI endoscopic examinations. We find close agreement between predicted labels using our method and the ground truth labelled by human experts. We observe the limitation of current static image classification scheme for EGD image classification. Conclusion: Our study presents a framework for developing automated EGD reports using deep learning. We demonstrate that our method is feasible to address EGD image classification and can lead towards improved performance and additionally qualitatively demonstrate its performance on our dataset.

Flexible manipulators offer significant advantages over traditional rigid manipulators in minimally invasive surgery, because they can flexibly navigate around obstacles and pass cramped or tortuous paths. However, due to the inherent low stiffness, the ability to control/obtain higher stiffness when required remains to be further explored. In this article, we propose a flexible manipulator that exploits the phase transformation property of low-melting-point alloy to hydraulically drive and change the stiffness by heating and cooling. A prototype was fabricated, and experiments were conducted to evaluate the motion characteristics, stiffness performance, and rigid-flexible transition efficiency. The experimental results demonstrate that the proposed manipulator can freely adjust heading direction in the three-dimensional space. The experimental results also indicate that it took 9.2–10.3 s for the manipulator to transform from a rigid state to a flexible state and 15.4 s to transform from a flexible state to a rigid state. The lateral stiffness and flexural stiffness of the manipulator were 95.54 and 372.1 Ncm ² in the rigid state and 7.26 and 0.78 Ncm ² in the flexible state. The gain of the lateral stiffness and flexural stiffness was 13.15 and 477.05, respectively. In the rigid state, the ultimate force without shape deformation was more than 0.98 N in the straight condition (0°) and 1.36 N in the bending condition (90°). By assembling flexible surgical tools, the manipulator can enrich the diagnosis or treatment functions, which demonstrated the potential clinical value of the proposed manipulator.