Radar-based gesture recognition

The proposed Diploma thesis will investigate the development of advanced radar signal processing and classification techniques to enable robust gesture recognition in real-world environments. Emphasis is placed on leveraging machine learning methods to extract discriminative spatio-temporal features from radar signatures associated with human gestures. To enhance annotation efficiency, the thesis will explore multimodal sensor fusion, integrating radar data with complementary modalities such as optical imaging. The resulting platform aims to support automated annotation pipelines, thereby reducing the dependency on manual data labeling. The proposed system will be validated through representative use cases, such as in-cabin gesture control for automotive infotainment systems, ensuring high recognition accuracy under varying environmental and operational conditions.