In this study, we present a new method for acquiring human vital signs using a Range-Doppler matrix (RDM) of FMCW radar data and a Gaussian interpolation algorithm (GIA). First, the RDM is derived by applying a two-dimensional fast Fourier transform (2D-FFT) to the radar data, and the GIA is applied in the Doppler dimension to estimate the target velocity signal. Subsequently, a robust enhanced trend filtering (RETF) algorithm is used to eliminate the large-scale body motion from the vital signs. Finally, the time-varying filter-based empirical mode decomposition (TVF-EMD) algorithm is employed to extract the respiratory and heartbeat intrinsic mode functions (IMFs), which are filtered according to their respective spectral power to obtain the respiratory and heartbeat frequencies. The proposed method was evaluated using vital signs data collected from seven volunteers (4 males and 3 females) with Texas Instrument’s AWR1642, and the results were compared with data from a reference monitor. The experiments showed that the method had an accuracy of 93 % for respiration and 95 % for heart rate in the presence of random body movements. Unlike traditional radar-based vital signs detection methods, this approach does not rely on range bin selection of the range profile matrix (RPM), thereby avoiding phase wrap problems and producing more accurate results. Currently, research in this field is limited.

Traditional path planning algorithms have drawbacks in unmanned vehicle path planning, such as low efficiency, unsuitability for dynamic environments, and high computational complexity. To address these issues, this paper proposes a dynamic path planning method for unmanned vehicles that combines the A* algorithm with an improved Dynamic Window Approach (DWA). Firstly, a collision risk function is incorporated into the DWA algorithm to evaluate scores based on the braking distance corresponding to the current vehicle speed and the distance to obstacles, determining whether timely braking or obstacle avoidance is feasible. The earlier the braking or obstacle avoidance, the higher the score, thereby enhancing the evaluation algorithm of the DWA. Secondly, the key points extracted by the A* algorithm are used as temporary target points for the DWA algorithm, reducing the eight search directions to five. Finally, the key points of the A* algorithm’s path are set as the temporary end points for the DWA algorithm, integrating the two algorithms to plan a smoother curved path that can avoid unknown and dynamic obstacles. Simulation data demonstrates that compared to traditional artificial potential field methods, RRT, and Dijkstra’s algorithm, the new fusion algorithm reduces time by 11.71 seconds, a decrease of 6.9%, and shortens path length by 3.093 meters, a reduction of 4.16%, while increasing average linear velocity by 2.9%. The proposed algorithm ensures smoother paths while also making paths more rational and efficient