An Improved PointLSTM Gesture Recognition Method Based on Millimeter-Wave Radar 3D Point Cloud

This paper aims to utilize millimeter-wave radar three-dimensional point cloud information for gesture recognition and proposes a novel PointLSTM network based on an external attention mechanism. By collecting echo signals of hand movements using a millimeter-wave radar chip with MIMO (Multiple Input Multiple Output) technology, and integrating it with a microcontroller equipped with a radar baseband processor, the study achieves real-time conversion from raw data to three-dimensional point clouds, significantly enhancing the processing speed of point clouds and the performance of radar hardware. The proposed network structure effectively addresses the deficiency of PointLSTM in inter-frame point information transmission and optimizes the complexity and recognition accuracy of the network through the external attention mechanism. Experimental results indicate that the proposed method achieves a detection accuracy rate of up to 98.3% in indoor environments, capable of accurately distinguishing different hand gestures, demonstrating the application potential of millimeter-wave radar three-dimensional point clouds in the field of gesture recognition. This research provides new insights and methods for the development of gesture recognition technology.(Abstract)