Localization of anti-personnel land mines using computationally modeled data for bistatic ground-coupled ground penetrating radar

Anti-personnel land mines are typically buried below a rough surface where the effectiveness of conventional air-coupled GPR is compromised. This work proposes that antennas be integrated onto the feet of a non-articulated walking robotic platform, in order to achieve ground-contact and thereby improve signal penetration, reduce rough ground clutter, and simplify data analysis. Using three antennas in both the transmitting and receiving modes, three unique bistatic GPR traces can be obtained, from which a target can be detected and localized geometrically using the known antenna locations and the full-path travel times. Using a 3-dimensional finite-difference time domain (FDTD) model, GPR signals are simulated for sixteen statistically different rough surfaces, and both metallic and non-metallic target casings. The soil modeled is both lossy and dispersive. Ultimately, this work demonstrates a detection and localization rate of 100%, independent of surface parameters and target casing, when a target is centered below the robotic platform.