Improved prediction of dismount velocity for tracking

The velocity measurements on moving objects like humans are complicated by the characteristics of bipedal or quadrupedal gait, in contrast to the bulk motion of a vehicle. We utilize the characteristic bipedal motion to recognize and track humans and vehicles. We characterize the moving objects based on its unpredictability and attempt to identify the class. We extract the velocity of the moving human and separate it from the micro-Doppler of the human motion in order to improve the speed of the velocity measurement and thus the extracted acceleration. We describe the detection, tracking, and characterization of moving objects like people and vehicles, as well as the radar sensors used for the measurements, and detail the velocity extraction at millisecond speeds. We discuss the robustness of the approach and potential improvements.