Advanced radar signal processing and spatial perception techniques for automotive, human-machine interaction, and remote sensing applications

This habilitation thesis presents advanced radar signal processing and spatial perception techniques developed to address key challenges in modern automated systems. The work spans fundamental research to applied innovations, enabling reliable and high-resolution perception across diverse domains such as automotive sensing, human–machine interac- tion (HMI), and interference-prone environments. The thesis introduces novel methods for high-resolution radar processing, including parametric and subspace-based algorithms for target separation in dense and complex scenes. InHMIapplications, radar-basedhandgesturerecognitionsystemsaredeveloped that leverage micro-Doppler signatures, multi-feature encoding, and edge-computing compatibility. To ensure robust radar operation in congested environments, several interference mitigation strategies are proposed, ranging from sparse sensing and deep learning to wavelet-domain morphological analysis. Cross-modal radar signal processing is explored through radar–camera fusion and automatic labeling techniques, improving semantic understanding and training efficiency for machine learning models. The thesis also investigates innovative Stepped-frequency PMCW (SF-PMCW) radar waveforms, offering high range resolution while avoiding stringent analog-digital-converter (ADC) requirements and motion-induced dispersion. Five granted patents verify the technical maturity of the developed methods in high- resolution processing and cross-modal fusion. The author’s involvement in the German federal project nxtAIM, in cooperation with Prof. Markus Gardill at Brandenburg Uni- versity of Technology Cottbus–Senftenberg, extends his research to generative AI-based radar perception. Recent work about the application of generative adversarial net- work (GAN) for range-Doppler imaging from one-bit phase-modulated continuous wave (PMCW) radar has been submitted to EUSIPCO 2025. Overall, this thesis demonstrates the versatile role of radar in advancing automated and intelligent systems, offering solutions that are both theoretically sound and practi- cally viable under real-world constraints.