The railroad industry needs to inspect and maintain the timber bridge on a regular basis to maximize its life. The timber deterioration can be minimized by inspecting, identifying, and recording information on the structure’s condition and performance so that timely corrective action can be taken to avoid extensive repair later or even replacement. However, accurately assessing the condition of a timber bridge structure is not easy. The innovation proposed in this project is expected to make this inspection easier, faster, and safer without compromising accuracy. The project will use a multi-sensor drone-based imaging synthetic aperture radar (iSAR) system to provide inspection of internal and external conditions of wooden railway bridge pilings. iSAR is a combination of a radar hardware design/data collection scheme paired with an advanced signal processing method which creates a focused 3D image from the raw radar data. The radar signal is capable of penetrating into the wooden pilings to provide internal condition information, which is not achievable with optical sensors. The image reconstruction feature improves the spatial resolution of the raw radar signals. A standard commercial drone, such as the DJI Matrice, will be outfitted with a custom sensor package which will include optical cameras, an IR camera, and an SAR array with supporting sensors such as real time tacking global positioning system (RTK GPS) and IMU (inertial measurement unit) for tracking position and a subsystem for collecting and storing the data. Operators will be able to scan a structure using a predefined flight plan (automated inspection) or by directly controlling flight path and data collection. During data collection, data from the various systems will be stored locally, which will include raw radar signals, camera imagery, and position information from the GPS/IMU. The data will then be uploaded to a remote cloud server and processed automatically. Data processing and analysis will involve data intake and verification (QA/QC of data streams), position data validation/correction, image reconstruction, feature extraction, feature correlations/data fusion and automated calculation of the piling quality metric. Raw data from different data streams and classification results will be then presented in a user-friendly secured geographic information system (GIS)-based web-enabled portal.

A complete commercialization-ready product will comprise a robust hardware solution and an accompanying software package which will include automated processing, classification algorithms, back-end (data handling/storage/lifecycle) and front-end (user experience) software needed to support integration with existing asset management/GIS software packages. The work will be accomplished in collaboration with the railroad industry partner, CSX.