Dual Stream Long Short-Term Memory Feature Fusion Classifier for Surface Electromyography Gesture Recognition.

Gesture recognition using electromyography (EMG) signals has prevailed recently in the field of human-computer interactions for controlling intelligent prosthetics. Currently, machine learning and deep learning are the two most commonly employed methods for classifying hand gestures. Despite traditional machine learning methods already achieving impressive performance, it is still a huge amount of work to carry out feature extraction manually. The existing deep learning methods utilize complex neural network architectures to achieve higher accuracy, which will suffer from overfitting, insufficient adaptability, and low recognition accuracy. To improve the existing phenomenon, a novel lightweight model named dual stream LSTM feature fusion classifier is proposed based on the concatenation of five time-domain features of EMG signals and raw data, which are both processed with one-dimensional convolutional neural networks and LSTM layers to carry out the classification. The proposed method can effectively capture global features of EMG signals using a simple architecture, which means less computational cost. An experiment is conducted on a public DB1 dataset with 52 gestures, and each of the 27 subjects repeats every gesture 10 times. The accuracy rate achieved by the model is 89.66%, which is comparable to that achieved by more complex deep learning neural networks, and the inference time for each gesture is 87.6 ms, which can also be implied in a real-time control system. The proposed model is validated using a subject-wise experiment on 10 out of the 40 subjects in the DB2 dataset, achieving a mean accuracy of 91.74%. This is illustrated by its ability to fuse time-domain features and raw data to extract more effective information from the sEMG signal and select an appropriate, efficient, lightweight network to enhance the recognition results.

(MARGOT) Monocular Camera-Based Robot Grasping Strategy for Metallic Objects.

Robotic handling of objects is not always a trivial assignment, even in teleoperation where, in most cases, this might lead to stressful labor for operators. To reduce the task difficulty, supervised motions could be performed in safe scenarios to reduce the workload in these non-critical steps by using machine learning and computer vision techniques. This paper describes a novel grasping strategy based on a groundbreaking geometrical analysis which extracts diametrically opposite points taking into account surface smoothing (even those target objects that might conform highly complex shapes) to guarantee the uniformity of the grasping. It uses a monocular camera, as we are often facing space restrictions that generate the need to use laparoscopic cameras integrated in the tools, to recognize and isolate targets from the background, estimating their spatial coordinates and providing the best possible stable grasping points for both feature and featureless objects. It copes with reflections and shadows produced by light sources (which require extra effort to extract their geometrical properties) in unstructured facilities such as nuclear power plants or particle accelerators on scientific equipment. Based on the experimental results, utilizing a specialized dataset improved the detection of metallic objects in low-contrast environments, resulting in the successful application of the algorithm with error rates in the scale of millimeters in the majority of repeatability and accuracy tests.