sEMG and NCLA-Based Gesture Recognition for Sewer Inspection Robot

In the domain of human–computer interaction (HCI), the recognition of emergency gestures based on surface electromyography (sEMG) signals is critical for minimizing the risk of inaccurate control in sewer inspection robots. This study is dedicated to establish a mapping relationship between forearm multichannel sEMG signals and emergency gestures, leading to the creation of the rainwater and sewage management gesture dataset (RSMGD) alongside a corresponding gesture recognition methodology. A comprehensive evaluation encompassing classification accuracy, CPU running time, and required feature dimensionality is conducted for this gesture recognition. Initially, RSMGD is collected utilizing Noraxon equipment, and an effective wavelet transform technique is applied to extract 2-D feature maps from the signals. To overcome the limitations of most traditional feature selection algorithms, which rely on a single fitness evaluation and involve high-dimensional features, this study introduces a novel feature optimization algorithm—the Nifty crow learning algorithm (NCLA). Inspired by the Lévy flight random walk model, originally used to simulate the stochastic movement and long-distance migratory behaviors of birds, the algorithm incorporates an innovative mutation strategy through crow following behavior and a memory updating mechanism, combined with a multitiered fitness evaluation mechanism, achieving more optimal feature selection. The results indicate that NCLA, using only 16 features (<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${p} \; \lt 0.05$ </tex-math></inline-formula>), achieves a classification accuracy of 99.04% in just 0.201 s (<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${p} \; \lt 0.05$ </tex-math></inline-formula>), with the true positive rate (TPR) and false positive rate (FPR) reaching 99.09% and 1.81%, respectively, demonstrating its exceptional performance in rapid and accurate gesture recognition.