RockSAR

About

Market Problem - The stability of a rock mass is governed by the geo-mechanical properties of its structural discontinuities. In underground mines, blasting endangers the frontline workers to an exposed rock face. This face might be unstable due to geological anomalies as well as blast induced discontinuities in the rock. A rock mass’s stability depends upon several factors such as the orientation, aperture, spacing, persistence, roughness, water flow as well as presence of filling material in the discontinuities. Apart from the existing methods of manually inspecting rock infrastructure, existing state-of-the-art and non-contact inspection techniques include use of ground penetrating radars (GPR), interferometric radars or LiDARs. Each of these techniques pose shortcomings which limit their use for surface discontinuity mapping. For example, a GPR provides the sub-surface profile beneath the rock surface. Most GPRs operate in the 10MHz – 2.6 GHz band of the RF spectrum, thus limiting their resolution to image surface features accurately. Similarly, interferometric radars observe the displacement maps of the surface under test. Thus, only after a certain change in displacement can they identify a risk zone. LiDARs have become extremely popular these days due to their capability to map a surface accurately using dense point clouds. Yet, LiDAR maps cannot detect presence of di-electric filling material or bands (shale bands, thin-bedded limestone, slaty lithologies) with sufficient contrast. Our Solution - RockSAR is a unique solution which provides high resolution imaging capability of surface discontinuities. It uses a millimeter-wave (mmWave) radar to scan the surface under test. With the help of a novel mmWave Synthetic Aperture Radar (SAR) imaging algorithm, it generates a high-fidelity radar image. Millimeter waves are extremely sensitive to surface aberrations. They can be used to detect tight aperture joints with sub-mm accuracy. Furthermore, the radar reflectivity depends upon the di-electric properties of the material. This means, the image reconstructed from this radar will provide the necessary contrast to filling materials, bands, and/or water flow. Additionally, RockSAR leverages a multi-modal imaging approach as it uses a LiDAR sensor to fuse its surface map with the reflectivity surface profile of the mmWave radar. The advantage of sensor fusion enables generation of radar reflectivity heat-maps co-registered with highly accurate 3D surface reconstruction of the LiDAR. Moreover, RockSAR makes use of Edge-AI that enables it to predict zone-wise geological confidence score from the images. This is made possible by the help of our patented edge module, a portable, wireless compute node that enables cloud-like computation capability at the edge of the network and in-situ AI inference. The same can interact, if required, with multiple RockSAR units in its range simultaneously providing domain-specific insights in the operator’s display module. Rather than depending upon displacement maps for warning generation, the AI engine localizes surface discontinuities, estimates the joint parameters, and predicts a site-specific geological confidence score. Thus, before any ground displacement takes place, the locations most susceptible to rock fall can be predicted and necessary measures to provide support system can be taken. RockSAR is compact and will be coupled with rocker-bogie ground rovers or drones to inspect in-accessible areas remotely. It is a one-stop solution which combines a physics driven sensing approach with geo-mechanics domain knowledge for robust and reliable rock infrastructure inspection. How it solves - The end-to-end RockSAR system comprises of three key units – RockSAR scanner, edge module and display module. The RockSAR scanner is a portable and compact sensor node housing a mmWave radar, a LiDAR sensor as well as a scanning rail arrangement. Using multi-sensor fusion, radar images are co-registered with the 3D surface generated from LiDAR scans. This provides enhanced 2D/3D visualization of the face/hangwall surface and its structural discontinuities to the operator. The scanner can be used standalone on a tripod or mounted upon a rover/UAV depending upon the degree of inaccessibility of the inspection site. For operation in development galleries and tunnels, the scanner can be used standalone mounted on top of a tripod stand. For remote inspection in raises or stopes, RockSAR can be coupled with an UGV/UAV, if required. The edge module is yet another portable, wireless unit which can be stationed anywhere within the range of the scanner. It utilizes a patented edge computing framework for in-situ execution of AI models. The display module comprises of a handy, rugged tablet with an operator friendly user interface that outputs 2D/3D surface visualization and interpretable insights of the AI engine in real-time.