Mobile Vibratory Impactor for AI-Based Detection of Void under AM-2 Matting

A mobile vibratory roller impactor system is proposed, using AI-based pattern recognition for identification of voids through analyzing the tapping force on the AM-2 mat exerted by the roller impactor. The Phase I project will conduct feasibility demonstration of the proposed technology. The basic concepts of the proposed technology will be implemented and demonstrated on a test pad or subgrade covered by AM-2 mat for detection of various voids embedded in the subgrade at different depths. AI-based pattern recognition techniques will be applied to locate the voids by identification of the features on the records of dynamic tapping force over them. The work will address some key issues in the technology development such as selection of magnitude of tapping impact, frequency, noise reduction, transducer selection and detection criteria development. The Phase I work will be divided in two sections: base effort for experimental feasibility demonstration and optional work for numerical simulation to provide guidance for further prototype development. Successful accomplishment of the Phase I and Phase I optional work will provide us with a solid background to proceed into the full scale prototype development.