Endovascular techniques have been increasingly adapted to medical applications as a minimally-invasive treatment approach to diagnose and treat various vascular diseases. Guidewires are the basic devices in endovascular surgery. The conventional practice of endovascular procedures has a number of drawbacks such as prolonged exposure to high doses of X-ray radiation, long operation time and undesired damage by surgeon’s unskilled operation. In this letter, we proposed a magnetically steerable guidewire and designed a three-dimensional magnetically actuating system. The proposed magnetically steerable guidewire consists of a commercial guidewire connecting to a soft actuator made of Ecoflex and magnetic powder. The actuating system contains two symmetrical parts, each consisting of three linear motors. It has great workspace of 700 mm × 500 mm × 500 mm, which can help the proposed guidewire to achieve a wide range of deflection angle. The relationship between the proposed guidewire’s deflection angle and the magnitude of the external magnetic field are explored, as well as the direction of the magnetic field. The proposed guidewire is flexible enough to make sharp turns up to 180 ${}^{\circ }$ in three dimensional workspace. The guidewire has achieved navigating through a set of rings and in a complex and constrained vascular phantom by programmed actuating magnetic fields.

In the lower limb exoskeleton assistance system, motion intention understanding based on biological signals is the human-computer interface’s (HMI) key technology. Due to the close coupling of human and machine, wheelchair-type control modes (combination of motion intentions and switches of different actions) are not applicable. For example, the right/left-hand motor imagery (MI) controls the exoskeleton to move forward/stop. There is a mismatch between MI-based control instructions and actual actions because the forward process alternates between left and right. Therefore, this study aims to design a reasonable HMI for a lower limb exoskeleton assistance system and to prove its rationality and effectiveness. We have designed a new HMI combining the human walking process and multimodal biosignals. Experimental data were collected from 14 subjects. We use feature-based and end-to-end analysis methods to analyze multimode signals. In a feature-based analysis, the advantage of using sEMG to distinguish between different movements does not apply to distinguishing different magnitudes of the same movement. A time-frequency subband selection method based on difference patterns is designed for EEG signals in the feature-based analysis. The fusion of EEG and sEMG is implemented to complement each other. The average recognition rate of the multimodal HMI was evaluated to be up to 89.5%, which is 12% and 7.8% higher than that of EEG and sEMG, respectively. Based on end-to-end analysis, a time-domain translation strategy was used to expand the sample and obtain a cross-subject identification rate of 83.6%. All the results indicate that the designed HMI is rational and effective, and there exists potential for practical application.

The two-inertia system with variable-length flexible load (TSVFL) is a typical dynamic model uncertain system. It is affected by transmission flexibility, flexible load length variation, and inaccurate friction torque, in which the flexible load will vibrate during rotation. In this study, a neural network compensation sliding mode control (SMC) strategy mixed with the angle-independent method (AIM) is proposed to suppress the vibration of the TSVFL. Among them, the AIM is proposed to design the fluctuating desired input of the motor. The vibration of the flexible load is offset by the speed fluctuation of the motor. The RBF neural networks are proposed to recognize the uncertain term in the TSVFL's dynamic model. First, the nonlinear mathematical model of the TSVFL is deduced based on the assumed mode method. Then, the Lyapunov stability theorem is proposed to design the weight coefficients’ adaptive law in neural networks and the robust term in the control law. Finally, simulation and physical experiments on output speed control are implemented. The results show that the proposed strategy is able to effectively decrease the error and weaken the vibration.

Gravity and flexibility will cause fluctuations of the rotation angle in the servo system for flexible manipulators. The fluctuation will seriously affect the motion accuracy of end-effectors. Therefore, this paper adopts a control method combining the RBF (Radial Basis Function) neural network and pole placement strategy to suppress the rotation angle fluctuations. The RBF neural network is used to identify uncertain items caused by the manipulator’s flexibility and the time-varying characteristics of dynamic parameters. Besides, the pole placement strategy is used to optimize the PD (Proportional Differential) controller’s parameters to improve the response speed and stability. Firstly, a dynamic model of flexible manipulators considering gravity is established based on the assumed mode method and Lagrange’s principle. Then, the system’s control characteristics are analyzed, and the pole placement strategy optimizes the parameters of the PD controllers. Next, the control method based on the RBF neural network is proposed, and the Lyapunov stability theory demonstrates stability. Finally, numerical analysis and control experiments prove the effectiveness of the control method proposed in this paper. The means and standard deviations of rotation angle error are reduced by the control method. The results show that the control method can effectively reduce the rotation angle error and improve motion accuracy.

Lateral resistance walk exercise (LRWE) is a popular method for fitness and rehabilitation training. However, current methods such as lateral band walks (LBW) cannot actively control the resistance training intensity. In this article, we proposed a novel hip exoskeleton, which can strengthen hip adductors by applying active resistance torque during lateral walking. The spatial linkage mechanism of the hip exoskeleton was designed and a prototype was fabricated. The dynamic model of transmission system coupling exoskeleton and human body was established. The proportional-integral-differential (PID) control strategy based on fuzzy tuning was presented to control the resistance torque. Physical prototype experiments showed that the fuzzy tuning PID control strategy could significantly improve the torque tracking accuracy compared to the traditional PID control strategy. The muscle activities of No-exo, Exo-off, LBW, and Exo-on (10 Nm, 15 Nm, 20 Nm) conditions were evaluated on ten healthy male subjects walking laterally at a speed of one step per second. The muscle activities of gluteus medius increased by 51.4%, 413.5%, 591.9%, 721.6%, and 918.9% under Exo-off, LBW, and Exo-on (10 Nm, 15 Nm, 20 Nm) conditions, respectively. The corresponding increments for tensor fasciae latae were 52.6%, 1136.8%, 1626.3%, 1994.7%, and 2331.6%, respectively. The results demonstrate that the proposed hip exoskeleton can apply to LRWE and improve muscle activities of hip adductors. It will upgrade the exercise method of LRWE and has good potential in strengthening hip abductors.

With the advances of robot technology in the medical field, robotic palpation has been a research hot topic and attracted wide attention. Tumor recognition is considered as a significant work in robotic palpation. Especially, the determination of tumor depth is a key prerequisite for tumor resection in robot-assisted minimally invasive surgery (RMIS). However, the lack of tactile feedback in RMIS prevents the robot from accurately perceiving information about lesions such as tumor depth. To solve this problem, we investigate the recognition of the depths of hard inclusion in soft tissue phantoms. In this article, tactile array data acquired by autonomous robotic palpation are used to recognize different depths of hard inclusions. Aiming at the classification problem of tumor depth recognition, we perform various deep learning methods and present ordinal classification methods based on the 2-D convolutional neural network- long short-term memory (2-D-CNN-LSTM) architecture for high-dimensional spatial–temporal tactile array data. The methods exploit the ordinal information of labels by changing the conventional one-hot encoding to unimodal probability distribution encodings, including Gaussian distribution, Cauchy distribution, and Laplace distribution encoding. Finally, the experimental results prove the feasibility of the proposed methods. The average recognition accuracies of the methods can reach 97%. In particular, the recognition accuracies can be above 99% when using all the robotic fingers’ tactile data.