In the industrial environment, Autonomous Guided Vehicles (AGVs) generally run on a planned route. Among trajectory-tracking algorithms for unmanned vehicles, the Pure Pursuit (PP) algorithm is prevalent in many real-world applications because of its simple and easy implementation. However, it is challenging to decelerate the AGV’s moving speed when turning on a large curve path. Moreover, this paper addresses the kidnapped-robot problem occurring in spare LiDAR environments. This paper proposes an improved Pure Pursuit algorithm so that the AGV can predict the trajectory and decelerate for turning, thus increasing the accuracy of the path tracking. To solve the kidnapped-robot problem, we use a learning-based classifier to detect the repetitive pattern scenario (e.g., long corridor) regarding 2D LiDAR features for switching the localization system between Simultaneous Localization And Mapping (SLAM) method and Odometer method. As experimental results in practice, the improved Pure Pursuit algorithm can reduce the tracking error while performing more efficiently. Moreover, the learning-based localization selection strategy helps the robot navigation task achieve stable performance, with 36.25% in completion rate more than only using SLAM. The results demonstrate that the proposed method is feasible and reliable in actual conditions.

Force detection is an important indicator of a robot's interaction with the working environment. Therefore, accurate and sensitive torque sensor design and development are important. This study proposed a robot joint torque sensor design and optimization based on a strain gauge and through-hole spoke structure topology to detect the external force with high sensitivity and low noise. A through-hole spoke type with a novel inclination sensing surface (ISS) was proposed, and the optimization design improved the strain gauge placement position and focused stress arrangement. Three optimization objectives were desired to enhance the joint torque sensor sensitivity with guaranteed stiffness. A sensor signal processing circuit was also introduced to obtain low-noise output signal performance. The experimental results showed that the joint torque sensor had a high sensitivity of 1.65 mV/Nm before amplification. This sensor was compared with a previous study and a commercial torque sensor. The prototype torque sensor indicated high sensitivity and low error.

Robotic cleaners are market-proven service robots. The most challenging issue of producing a robotic cleaner is to maximize its coverage performance. The complete coverage performance depends on the obstacle detection sensors as well as on the navigation systems. Conventional contact, infrared, and ultrasonic sensors perform either collision-based or collision-free perceptions with obstacles, and these sensors represent a poor coverage performance when the robotic cleaner moves beside the obstacle or wall. Therefore, this paper proposes a soft-contact-based pneumatic sensor with four closed chambers to continuously contact objects with a certain level of force when the robotic cleaner touches the obstacle or wall so that the complete coverage performance could be improved. Practically, a robotic cleaner navigation system is developed using the field programmable gate array. The finite state machine is further used to model the control strategy of the navigation systems according to the sensor data and elapsed time. Finally, several experiments were evaluated based on the path of navigation and the percentage of coverage with elapsed time to verify our approaches, and the improved complete coverage performance is verified by comparing the performance of iRobot Create using the same test environments and conditions.

This study introduces a novel frequency-based attention network (FANet) designed to acquire frequency information in surface electromyographic signals (sEMG) while mitigating potential high-frequency disturbances. The key innovation is the utilization of the discrete cosine transform (DCT) in dedicated attention modules, enabling the conversion of sEMG features to the frequency domain and facilitating the extraction of spectral characteristics. This attention module replaces the self-attention module within the multilayer perceptron (MLP) based transformer, offering a streamlined means for internal frequency features analysis in deep learning models. To evaluate the proposed models, we focus on the dominance of the “rest” gesture in available NinaPro DB4 and NinaPro DB5 datasets. Consequently, we employ common evaluation metrics tailored to the class-imbalanced hand gesture classification task. Through comprehensive experiments and ablation studies, the results confirm the effectiveness of the proposed network in capturing frequency insights, leading to substantial improvement in hand gesture recognition. Notably, the DCT-based attention modules achieve accuracies of 78.70% and 89.64% in the NinaPro DB4 and DB5 datasets, respectively, while maintaining slight computational cost with log-linear complexity.

This paper describes an electrooculography (EOG) based human-wheelchair interface for wheelchair users. A pair of electrodes which measures the eye-gaze direction of users is desired as wheelchair manipulation commands. In addition to EOG based wheelchair manipulations, this paper also introduces ultrasonic arrays for detecting distances to actively avoid collisions. In order to simplify setup procedures of using this system, two electrodes are mounted on two side arms of eyeglasses respectively that are possible to have tight contacts with skins around eyes. Therefore, wheelchair movements are capable of following eye-gaze directions. On the other hand, an analog biopotential signal amplifier and a laptop computer are used to develop the proposed EOG based human-wheelchair interface controller. Finally, experimental results demonstrate the operations of EOG based human-wheelchair interface.

This study introduces a novel approach for real-time hand gesture classification through the integration of a Frequency-based Attention Neural Network (FANN) with Subject-Adaptive Transfer Learning, specifically tailored for surface electromyography (sEMG) data. By utilizing the Fourier transform, the proposed methodology leverages the inherent frequency characteristics of sEMG signals to enhance the discriminative features for accurate gesture recognition. Additionally, the subject-adaptive transfer learning strategy is employed to improve model generalization across different individuals. The combination of these techniques results in an effective and versatile system for sEMG-based hand gesture classification, demonstrating promising performance in adapting individual variability and improving classification accuracy. The proposed method's performance is evaluated and compared with established approaches using the publicity available NinaPro DB5 dataset. Notably, the proposed simple model, coupled with frequency-based attention modules, achieves accuracies of 89.56% with a quick prediction time of 5ms, showcasing its potential for dexterous control of robots and bionic hands. The findings of this research contribute to the advancement of gesture recognition systems, particularly in the domains of human-computer interaction and prosthetic control.